Tag Archives: blocking

Some notes on eFail

Post Syndicated from Robert Graham original https://blog.erratasec.com/2018/05/some-notes-on-efail.html

I’ve been busy trying to replicate the “eFail” PGP/SMIME bug. I thought I’d write up some notes.

PGP and S/MIME encrypt emails, so that eavesdroppers can’t read them. The bugs potentially allow eavesdroppers to take the encrypted emails they’ve captured and resend them to you, reformatted in a way that allows them to decrypt the messages.

Disable remote/external content in email

The most important defense is to disable “external” or “remote” content from being automatically loaded. This is when HTML-formatted emails attempt to load images from remote websites. This happens legitimately when they want to display images, but not fill up the email with them. But most of the time this is illegitimate, they hide images on the webpage in order to track you with unique IDs and cookies. For example, this is the code at the end of an email from politician Bernie Sanders to his supporters. Notice the long random number assigned to track me, and the width/height of this image is set to one pixel, so you don’t even see it:

Such trackers are so pernicious they are disabled by default in most email clients. This is an example of the settings in Thunderbird:

The problem is that as you read email messages, you often get frustrated by the fact the error messages and missing content, so you keep adding exceptions:

The correct defense against this eFail bug is to make sure such remote content is disabled and that you have no exceptions, or at least, no HTTP exceptions. HTTPS exceptions (those using SSL) are okay as long as they aren’t to a website the attacker controls. Unencrypted exceptions, though, the hacker can eavesdrop on, so it doesn’t matter if they control the website the requests go to. If the attacker can eavesdrop on your emails, they can probably eavesdrop on your HTTP sessions as well.

Some have recommended disabling PGP and S/MIME completely. That’s probably overkill. As long as the attacker can’t use the “remote content” in emails, you are fine. Likewise, some have recommend disabling HTML completely. That’s not even an option in any email client I’ve used — you can disable sending HTML emails, but not receiving them. It’s sufficient to just disable grabbing remote content, not the rest of HTML email rendering.

I couldn’t replicate the direct exfiltration

There rare two related bugs. One allows direct exfiltration, which appends the decrypted PGP email onto the end of an IMG tag (like one of those tracking tags), allowing the entire message to be decrypted.

An example of this is the following email. This is a standard HTML email message consisting of multiple parts. The trick is that the IMG tag in the first part starts the URL (blog.robertgraham.com/…) but doesn’t end it. It has the starting quotes in front of the URL but no ending quotes. The ending will in the next chunk.

The next chunk isn’t HTML, though, it’s PGP. The PGP extension (in my case, Enignmail) will detect this and automatically decrypt it. In this case, it’s some previous email message I’ve received the attacker captured by eavesdropping, who then pastes the contents into this email message in order to get it decrypted.

What should happen at this point is that Thunderbird will generate a request (if “remote content” is enabled) to the blog.robertgraham.com server with the decrypted contents of the PGP email appended to it. But that’s not what happens. Instead, I get this:

I am indeed getting weird stuff in the URL (the bit after the GET /), but it’s not the PGP decrypted message. Instead what’s going on is that when Thunderbird puts together a “multipart/mixed” message, it adds it’s own HTML tags consisting of lines between each part. In the email client it looks like this:

The HTML code it adds looks like:

That’s what you see in the above URL, all this code up to the first quotes. Those quotes terminate the quotes in the URL from the first multipart section, causing the rest of the content to be ignored (as far as being sent as part of the URL).

So at least for the latest version of Thunderbird, you are accidentally safe, even if you have “remote content” enabled. Though, this is only according to my tests, there may be a work around to this that hackers could exploit.

STARTTLS

In the old days, email was sent plaintext over the wire so that it could be passively eavesdropped on. Nowadays, most providers send it via “STARTTLS”, which sorta encrypts it. Attackers can still intercept such email, but they have to do so actively, using man-in-the-middle. Such active techniques can be detected if you are careful and look for them.
Some organizations don’t care. Apparently, some nation states are just blocking all STARTTLS and forcing email to be sent unencrypted. Others do care. The NSA will passively sniff all the email they can in nations like Iraq, but they won’t actively intercept STARTTLS messages, for fear of getting caught.
The consequence is that it’s much less likely that somebody has been eavesdropping on you, passively grabbing all your PGP/SMIME emails. If you fear they have been, you should look (e.g. send emails from GMail and see if they are intercepted by sniffing the wire).

You’ll know if you are getting hacked

If somebody attacks you using eFail, you’ll know. You’ll get an email message formatted this way, with multipart/mixed components, some with corrupt HTML, some encrypted via PGP. This means that for the most part, your risk is that you’ll be attacked only once — the hacker will only be able to get one message through and decrypt it before you notice that something is amiss. Though to be fair, they can probably include all the emails they want decrypted as attachments to the single email they sent you, so the risk isn’t necessarily that you’ll only get one decrypted.
As mentioned above, a lot of attackers (e.g. the NSA) won’t attack you if its so easy to get caught. Other attackers, though, like anonymous hackers, don’t care.
Somebody ought to write a plugin to Thunderbird to detect this.

Summary

It only works if attackers have already captured your emails (though, that’s why you use PGP/SMIME in the first place, to guard against that).
It only works if you’ve enabled your email client to automatically grab external/remote content.
It seems to not be easily reproducible in all cases.
Instead of disabling PGP/SMIME, you should make sure your email client hast remote/external content disabled — that’s a huge privacy violation even without this bug.

Notes: The default email client on the Mac enables remote content by default, which is bad:

Russia is Banning Telegram

Post Syndicated from Bruce Schneier original https://www.schneier.com/blog/archives/2018/04/russia_is_banni.html

Russia has banned the secure messaging app Telegram. It’s making an absolute mess of the ban — blocking 16 million IP addresses, many belonging to the Amazon and Google clouds — and it’s not even clear that it’s working. But, more importantly, I’m not convinced Telegram is secure in the first place.

Such a weird story. If you want secure messaging, use Signal. If you’re concerned that having Signal on your phone will itself arouse suspicion, use WhatsApp.

User Authentication Best Practices Checklist

Post Syndicated from Bozho original https://techblog.bozho.net/user-authentication-best-practices-checklist/

User authentication is the functionality that every web application shared. We should have perfected that a long time ago, having implemented it so many times. And yet there are so many mistakes made all the time.

Part of the reason for that is that the list of things that can go wrong is long. You can store passwords incorrectly, you can have a vulnerably password reset functionality, you can expose your session to a CSRF attack, your session can be hijacked, etc. So I’ll try to compile a list of best practices regarding user authentication. OWASP top 10 is always something you should read, every year. But that might not be enough.

So, let’s start. I’ll try to be concise, but I’ll include as much of the related pitfalls as I can cover – e.g. what could go wrong with the user session after they login:

  • Store passwords with bcrypt/scrypt/PBKDF2. No MD5 or SHA, as they are not good for password storing. Long salt (per user) is mandatory (the aforementioned algorithms have it built in). If you don’t and someone gets hold of your database, they’ll be able to extract the passwords of all your users. And then try these passwords on other websites.
  • Use HTTPS. Period. (Otherwise user credentials can leak through unprotected networks). Force HTTPS if user opens a plain-text version.
  • Mark cookies as secure. Makes cookie theft harder.
  • Use CSRF protection (e.g. CSRF one-time tokens that are verified with each request). Frameworks have such functionality built-in.
  • Disallow framing (X-Frame-Options: DENY). Otherwise your website may be included in another website in a hidden iframe and “abused” through javascript.
  • Have a same-origin policy
  • Logout – let your users logout by deleting all cookies and invalidating the session. This makes usage of shared computers safer (yes, users should ideally use private browsing sessions, but not all of them are that savvy)
  • Session expiry – don’t have forever-lasting sessions. If the user closes your website, their session should expire after a while. “A while” may still be a big number depending on the service provided. For ajax-heavy website you can have regular ajax-polling that keeps the session alive while the page stays open.
  • Remember me – implementing “remember me” (on this machine) functionality is actually hard due to the risks of a stolen persistent cookie. Spring-security uses this approach, which I think should be followed if you wish to implement more persistent logins.
  • Forgotten password flow – the forgotten password flow should rely on sending a one-time (or expiring) link to the user and asking for a new password when it’s opened. 0Auth explain it in this post and Postmark gives some best pracitces. How the link is formed is a separate discussion and there are several approaches. Store a password-reset token in the user profile table and then send it as parameter in the link. Or do not store anything in the database, but send a few params: userId:expiresTimestamp:hmac(userId+expiresTimestamp). That way you have expiring links (rather than one-time links). The HMAC relies on a secret key, so the links can’t be spoofed. It seems there’s no consensus, as the OWASP guide has a bit different approach
  • One-time login links – this is an option used by Slack, which sends one-time login links instead of asking users for passwords. It relies on the fact that your email is well guarded and you have access to it all the time. If your service is not accessed to often, you can have that approach instead of (rather than in addition to) passwords.
  • Limit login attempts – brute-force through a web UI should not be possible; therefore you should block login attempts if they become too many. One approach is to just block them based on IP. The other one is to block them based on account attempted. (Spring example here). Which one is better – I don’t know. Both can actually be combined. Instead of fully blocking the attempts, you may add a captcha after, say, the 5th attempt. But don’t add the captcha for the first attempt – it is bad user experience.
  • Don’t leak information through error messages – you shouldn’t allow attackers to figure out if an email is registered or not. If an email is not found, upon login report just “Incorrect credentials”. On passwords reset, it may be something like “If your email is registered, you should have received a password reset email”. This is often at odds with usability – people don’t often remember the email they used to register, and the ability to check a number of them before getting in might be important. So this rule is not absolute, though it’s desirable, especially for more critical systems.
  • Make sure you use JWT only if it’s really necessary and be careful of the pitfalls.
  • Consider using a 3rd party authentication – OpenID Connect, OAuth by Google/Facebook/Twitter (but be careful with OAuth flaws as well). There’s an associated risk with relying on a 3rd party identity provider, and you still have to manage cookies, logout, etc., but some of the authentication aspects are simplified.
  • For high-risk or sensitive applications use 2-factor authentication. There’s a caveat with Google Authenticator though – if you lose your phone, you lose your accounts (unless there’s a manual process to restore it). That’s why Authy seems like a good solution for storing 2FA keys.

I’m sure I’m missing something. And you see it’s complicated. Sadly we’re still at the point where the most common functionality – authenticating users – is so tricky and cumbersome, that you almost always get at least some of it wrong.

The post User Authentication Best Practices Checklist appeared first on Bozho's tech blog.

Node.js 8.10 runtime now available in AWS Lambda

Post Syndicated from Chris Munns original https://aws.amazon.com/blogs/compute/node-js-8-10-runtime-now-available-in-aws-lambda/

This post courtesy of Ed Lima, AWS Solutions Architect

We are excited to announce that you can now develop your AWS Lambda functions using the Node.js 8.10 runtime, which is the current Long Term Support (LTS) version of Node.js. Start using this new version today by specifying a runtime parameter value of nodejs8.10 when creating or updating functions.

Supporting async/await

The Lambda programming model for Node.js 8.10 now supports defining a function handler using the async/await pattern.

Asynchronous or non-blocking calls are an inherent and important part of applications, as user and human interfaces are asynchronous by nature. If you decide to have a coffee with a friend, you usually order the coffee then start or continue a conversation with your friend while the coffee is getting ready. You don’t wait for the coffee to be ready before you start talking. These activities are asynchronous, because you can start one and then move to the next without waiting for completion. Otherwise, you’d delay (or block) the start of the next activity.

Asynchronous calls used to be handled in Node.js using callbacks. That presented problems when they were nested within other callbacks in multiple levels, making the code difficult to maintain and understand.

Promises were implemented to try to solve issues caused by “callback hell.” They allow asynchronous operations to call their own methods and handle what happens when a call is successful or when it fails. As your requirements become more complicated, even promises become harder to work with and may still end up complicating your code.

Async/await is the new way of handling asynchronous operations in Node.js, and makes for simpler, easier, and cleaner code for non-blocking calls. It still uses promises but a callback is returned directly from the asynchronous function, just as if it were a synchronous blocking function.

Take for instance the following Lambda function to get the current account settings, using the Node.js 6.10 runtime:

let AWS = require('aws-sdk');
let lambda = new AWS.Lambda();

exports.handler = (event, context, callback) => {
    let getAccountSettingsPromise = lambda.getAccountSettings().promise();
    getAccountSettingsPromise.then(
        (data) => {
            callback(null, data);
        },
        (err) => {
            console.log(err);
            callback(err);
        }
    );
};

With the new Node.js 8.10 runtime, there are new handler types that can be declared with the “async” keyword or can return a promise directly.

This is how the same function looks like using async/await with Node.js 8.10:

let AWS = require('aws-sdk');
let lambda = new AWS.Lambda();

exports.handler = async (event) => {
    return await lambda.getAccountSettings().promise() ;
};

Alternatively, you could have the handler return a promise directly:

let AWS = require('aws-sdk');
let lambda = new AWS.Lambda();

exports.handler = (event) => {
    return new Promise((resolve, reject) => {
        lambda.getAccountSettings(event)
        .then((data) => {
            resolve data;
        })
        .catch(reject);
     });
};

The new handler types are alternatives to the callback pattern, which is still fully supported.

All three functions return the same results. However, in the new runtime with async/await, all callbacks in the code are gone, which makes it easier to read. This is especially true for those less familiar with promises.

{
    "AccountLimit":{
        "TotalCodeSize":80530636800,
        "CodeSizeUnzipped":262144000,
        "CodeSizeZipped":52428800, 
        "ConcurrentExecutions":1000,
        "UnreservedConcurrentExecutions":1000
    },
    "AccountUsage":{
        "TotalCodeSize":52234461,
        "FunctionCount":53
    }
}

Another great advantage of async/await is better error handling. You can use a try/catch block inside the scope of an async function. Even though the function awaits an asynchronous operation, any errors end up in the catch block.

You can improve your previous Node.js 8.10 function with this trusted try/catch error handling pattern:

let AWS = require('aws-sdk');
let lambda = new AWS.Lambda();
let data;

exports.handler = async (event) => {
    try {
        data = await lambda.getAccountSettings().promise();
    }
    catch (err) {
        console.log(err);
        return err;
    }
    return data;
};

While you now have a similar number of lines in both runtimes, the code is cleaner and more readable with async/await. It makes the asynchronous calls look more synchronous. However, it is important to notice that the code is still executed the same way as if it were using a callback or promise-based API.

Backward compatibility

You may port your existing Node.js 4.3 and 6.10 functions over to Node.js 8.10 by updating the runtime. Node.js 8.10 does include numerous breaking changes from previous Node versions.

Make sure to review the API changes between Node.js 4.3, 6.10, and Node.js 8.10 to see if there are other changes that might affect your code. We recommend testing that your Lambda function passes internal validation for its behavior when upgrading to the new runtime version.

You can use Lambda versions/aliases to safely test that your function runs as expected on Node 8.10, before routing production traffic to it.

New node features

You can now get better performance when compared to the previous LTS version 6.x (up to 20%). The new V8 6.0 engine comes with Turbofan and the Ignition pipeline, which leads to lower memory consumption and faster startup time across Node.js applications.

HTTP/2, which is subject to future changes, allows developers to use the new protocol to speed application development and undo many of HTTP/1.1 workarounds to make applications faster, simpler, and more powerful.

For more information, see the AWS Lambda Developer Guide.

Hope you enjoy and… go build with Node.js 8.10!

[$] Recent improvements to Tor

Post Syndicated from jake original https://lwn.net/Articles/750312/rss

We may need Tor, “the onion router”,
more than we ever imagined. Authoritarian states are blocking more and more web
sites
and snooping
on their populations online
—even routine tracking of our online
activities
can reveal information that can be used to undermine
democracy. Thus, there was strong interest in the “State of the Onion”
panel at the 2018 LibrePlanet conference, where
four contributors to the Tor project presented a progress update covering the
past few years.

Subscribers can read on for a report on the panel by guest author Andy Oram.

Tech wishes for 2018

Post Syndicated from Eevee original https://eev.ee/blog/2018/02/18/tech-wishes-for-2018/

Anonymous asks, via money:

What would you like to see happen in tech in 2018?

(answer can be technical, social, political, combination, whatever)

Hmm.

Less of this

I’m not really qualified to speak in depth about either of these things, but let me put my foot in my mouth anyway:

The Blockchain™

Bitcoin was a neat idea. No, really! Decentralization is cool. Overhauling our terrible financial infrastructure is cool. Hash functions are cool.

Unfortunately, it seems to have devolved into mostly a get-rich-quick scheme for nerds, and by nearly any measure it’s turning into a spectacular catastrophe. Its “success” is measured in how much a bitcoin is worth in US dollars, which is pretty close to an admission from its own investors that its only value is in converting back to “real” money — all while that same “success” is making it less useful as a distinct currency.

Blah, blah, everyone already knows this.

What concerns me slightly more is the gold rush hype cycle, which is putting cryptocurrency and “blockchain” in the news and lending it all legitimacy. People have raked in millions of dollars on ICOs of novel coins I’ve never heard mentioned again. (Note: again, that value is measured in dollars.) Most likely, none of the investors will see any return whatsoever on that money. They can’t, really, unless a coin actually takes off as a currency, and that seems at odds with speculative investing since everyone either wants to hoard or ditch their coins. When the coins have no value themselves, the money can only come from other investors, and eventually the hype winds down and you run out of other investors.

I fear this will hurt a lot of people before it’s over, so I’d like for it to be over as soon as possible.


That said, the hype itself has gotten way out of hand too. First it was the obsession with “blockchain” like it’s a revolutionary technology, but hey, Git is a fucking blockchain. The novel part is the way it handles distributed consensus (which in Git is basically left for you to figure out), and that’s uniquely important to currency because you want to be pretty sure that money doesn’t get duplicated or lost when moved around.

But now we have startups trying to use blockchains for website backends and file storage and who knows what else? Why? What advantage does this have? When you say “blockchain”, I hear “single Git repository” — so when you say “email on the blockchain”, I have an aneurysm.

Bitcoin seems to have sparked imagination in large part because it’s decentralized, but I’d argue it’s actually a pretty bad example of a decentralized network, since people keep forking it. The ability to fork is a feature, sure, but the trouble here is that the Bitcoin family has no notion of federation — there is one canonical Bitcoin ledger and it has no notion of communication with any other. That’s what you want for currency, not necessarily other applications. (Bitcoin also incentivizes frivolous forking by giving the creator an initial pile of coins to keep and sell.)

And federation is much more interesting than decentralization! Federation gives us email and the web. Federation means I can set up my own instance with my own rules and still be able to meaningfully communicate with the rest of the network. Federation has some amount of tolerance for changes to the protocol, so such changes are more flexible and rely more heavily on consensus.

Federation is fantastic, and it feels like a massive tragedy that this rekindled interest in decentralization is mostly focused on peer-to-peer networks, which do little to address our current problems with centralized platforms.

And hey, you know what else is federated? Banks.

AI

Again, the tech is cool and all, but the marketing hype is getting way out of hand.

Maybe what I really want from 2018 is less marketing?

For one, I’ve seen a huge uptick in uncritically referring to any software that creates or classifies creative work as “AI”. Can we… can we not. It’s not AI. Yes, yes, nerds, I don’t care about the hair-splitting about the nature of intelligence — you know that when we hear “AI” we think of a human-like self-aware intelligence. But we’re applying it to stuff like a weird dog generator. Or to whatever neural network a website threw into production this week.

And this is dangerously misleading — we already had massive tech companies scapegoating The Algorithm™ for the poor behavior of their software, and now we’re talking about those algorithms as though they were self-aware, untouchable, untameable, unknowable entities of pure chaos whose decisions we are arbitrarily bound to. Ancient, powerful gods who exist just outside human comprehension or law.

It’s weird to see this stuff appear in consumer products so quickly, too. It feels quick, anyway. The latest iPhone can unlock via facial recognition, right? I’m sure a lot of effort was put into ensuring that the same person’s face would always be recognized… but how confident are we that other faces won’t be recognized? I admit I don’t follow all this super closely, so I may be imagining a non-problem, but I do know that humans are remarkably bad at checking for negative cases.

Hell, take the recurring problem of major platforms like Twitter and YouTube classifying anything mentioning “bisexual” as pornographic — because the word is also used as a porn genre, and someone threw a list of porn terms into a filter without thinking too hard about it. That’s just a word list, a fairly simple thing that any human can review; but suddenly we’re confident in opaque networks of inferred details?

I don’t know. “Traditional” classification and generation are much more comforting, since they’re a set of fairly abstract rules that can be examined and followed. Machine learning, as I understand it, is less about rules and much more about pattern-matching; it’s built out of the fingerprints of the stuff it’s trained on. Surely that’s just begging for tons of edge cases. They’re practically made of edge cases.


I’m reminded of a point I saw made a few days ago on Twitter, something I’d never thought about but should have. TurnItIn is a service for universities that checks whether students’ papers match any others, in order to detect cheating. But this is a paid service, one that fundamentally hinges on its corpus: a large collection of existing student papers. So students pay money to attend school, where they’re required to let their work be given to a third-party company, which then profits off of it? What kind of a goofy business model is this?

And my thoughts turn to machine learning, which is fundamentally different from an algorithm you can simply copy from a paper, because it’s all about the training data. And to get good results, you need a lot of training data. Where is that all coming from? How many for-profit companies are setting a neural network loose on the web — on millions of people’s work — and then turning around and selling the result as a product?

This is really a question of how intellectual property works in the internet era, and it continues our proud decades-long tradition of just kinda doing whatever we want without thinking about it too much. Nothing if not consistent.

More of this

A bit tougher, since computers are pretty alright now and everything continues to chug along. Maybe we should just quit while we’re ahead. There’s some real pie-in-the-sky stuff that would be nice, but it certainly won’t happen within a year, and may never happen except in some horrific Algorithmic™ form designed by people that don’t know anything about the problem space and only works 60% of the time but is treated as though it were bulletproof.

Federation

The giants are getting more giant. Maybe too giant? Granted, it could be much worse than Google and Amazon — it could be Apple!

Amazon has its own delivery service and brick-and-mortar stores now, as well as providing the plumbing for vast amounts of the web. They’re not doing anything particularly outrageous, but they kind of loom.

Ad company Google just put ad blocking in its majority-share browser — albeit for the ambiguously-noble goal of only blocking obnoxious ads so that people will be less inclined to install a blanket ad blocker.

Twitter is kind of a nightmare but no one wants to leave. I keep trying to use Mastodon as well, but I always forget about it after a day, whoops.

Facebook sounds like a total nightmare but no one wants to leave that either, because normies don’t use anything else, which is itself direly concerning.

IRC is rapidly bleeding mindshare to Slack and Discord, both of which are far better at the things IRC sadly never tried to do and absolutely terrible at the exact things IRC excels at.

The problem is the same as ever: there’s no incentive to interoperate. There’s no fundamental technical reason why Twitter and Tumblr and MySpace and Facebook can’t intermingle their posts; they just don’t, because why would they bother? It’s extra work that makes it easier for people to not use your ecosystem.

I don’t know what can be done about that, except that hope for a really big player to decide to play nice out of the kindness of their heart. The really big federated success stories — say, the web — mostly won out because they came along first. At this point, how does a federated social network take over? I don’t know.

Social progress

I… don’t really have a solid grasp on what’s happening in tech socially at the moment. I’ve drifted a bit away from the industry part, which is where that all tends to come up. I have the vague sense that things are improving, but that might just be because the Rust community is the one I hear the most about, and it puts a lot of effort into being inclusive and welcoming.

So… more projects should be like Rust? Do whatever Rust is doing? And not so much what Linus is doing.

Open source funding

I haven’t heard this brought up much lately, but it would still be nice to see. The Bay Area runs on open source and is raking in zillions of dollars on its back; pump some of that cash back into the ecosystem, somehow.

I’ve seen a couple open source projects on Patreon, which is fantastic, but feels like a very small solution given how much money is flowing through the commercial tech industry.

Ad blocking

Nice. Fuck ads.

One might wonder where the money to host a website comes from, then? I don’t know. Maybe we should loop this in with the above thing and find a more informal way to pay people for the stuff they make when we find it useful, without the financial and cognitive overhead of A Transaction or Giving Someone My Damn Credit Card Number. You know, something like Bitco— ah, fuck.

Year of the Linux Desktop

I don’t know. What are we working on at the moment? Wayland? Do Wayland, I guess. Oh, and hi-DPI, which I hear sucks. And please fix my sound drivers so PulseAudio stops blaming them when it fucks up.

Reactive Microservices Architecture on AWS

Post Syndicated from Sascha Moellering original https://aws.amazon.com/blogs/architecture/reactive-microservices-architecture-on-aws/

Microservice-application requirements have changed dramatically in recent years. These days, applications operate with petabytes of data, need almost 100% uptime, and end users expect sub-second response times. Typical N-tier applications can’t deliver on these requirements.

Reactive Manifesto, published in 2014, describes the essential characteristics of reactive systems including: responsiveness, resiliency, elasticity, and being message driven.

Being message driven is perhaps the most important characteristic of reactive systems. Asynchronous messaging helps in the design of loosely coupled systems, which is a key factor for scalability. In order to build a highly decoupled system, it is important to isolate services from each other. As already described, isolation is an important aspect of the microservices pattern. Indeed, reactive systems and microservices are a natural fit.

Implemented Use Case
This reference architecture illustrates a typical ad-tracking implementation.

Many ad-tracking companies collect massive amounts of data in near-real-time. In many cases, these workloads are very spiky and heavily depend on the success of the ad-tech companies’ customers. Typically, an ad-tracking-data use case can be separated into a real-time part and a non-real-time part. In the real-time part, it is important to collect data as fast as possible and ask several questions including:,  “Is this a valid combination of parameters?,””Does this program exist?,” “Is this program still valid?”

Because response time has a huge impact on conversion rate in advertising, it is important for advertisers to respond as fast as possible. This information should be kept in memory to reduce communication overhead with the caching infrastructure. The tracking application itself should be as lightweight and scalable as possible. For example, the application shouldn’t have any shared mutable state and it should use reactive paradigms. In our implementation, one main application is responsible for this real-time part. It collects and validates data, responds to the client as fast as possible, and asynchronously sends events to backend systems.

The non-real-time part of the application consumes the generated events and persists them in a NoSQL database. In a typical tracking implementation, clicks, cookie information, and transactions are matched asynchronously and persisted in a data store. The matching part is not implemented in this reference architecture. Many ad-tech architectures use frameworks like Hadoop for the matching implementation.

The system can be logically divided into the data collection partand the core data updatepart. The data collection part is responsible for collecting, validating, and persisting the data. In the core data update part, the data that is used for validation gets updated and all subscribers are notified of new data.

Components and Services

Main Application
The main application is implemented using Java 8 and uses Vert.x as the main framework. Vert.x is an event-driven, reactive, non-blocking, polyglot framework to implement microservices. It runs on the Java virtual machine (JVM) by using the low-level IO library Netty. You can write applications in Java, JavaScript, Groovy, Ruby, Kotlin, Scala, and Ceylon. The framework offers a simple and scalable actor-like concurrency model. Vert.x calls handlers by using a thread known as an event loop. To use this model, you have to write code known as “verticles.” Verticles share certain similarities with actors in the actor model. To use them, you have to implement the verticle interface. Verticles communicate with each other by generating messages in  a single event bus. Those messages are sent on the event bus to a specific address, and verticles can register to this address by using handlers.

With only a few exceptions, none of the APIs in Vert.x block the calling thread. Similar to Node.js, Vert.x uses the reactor pattern. However, in contrast to Node.js, Vert.x uses several event loops. Unfortunately, not all APIs in the Java ecosystem are written asynchronously, for example, the JDBC API. Vert.x offers a possibility to run this, blocking APIs without blocking the event loop. These special verticles are called worker verticles. You don’t execute worker verticles by using the standard Vert.x event loops, but by using a dedicated thread from a worker pool. This way, the worker verticles don’t block the event loop.

Our application consists of five different verticles covering different aspects of the business logic. The main entry point for our application is the HttpVerticle, which exposes an HTTP-endpoint to consume HTTP-requests and for proper health checking. Data from HTTP requests such as parameters and user-agent information are collected and transformed into a JSON message. In order to validate the input data (to ensure that the program exists and is still valid), the message is sent to the CacheVerticle.

This verticle implements an LRU-cache with a TTL of 10 minutes and a capacity of 100,000 entries. Instead of adding additional functionality to a standard JDK map implementation, we use Google Guava, which has all the features we need. If the data is not in the L1 cache, the message is sent to the RedisVerticle. This verticle is responsible for data residing in Amazon ElastiCache and uses the Vert.x-redis-client to read data from Redis. In our example, Redis is the central data store. However, in a typical production implementation, Redis would just be the L2 cache with a central data store like Amazon DynamoDB. One of the most important paradigms of a reactive system is to switch from a pull- to a push-based model. To achieve this and reduce network overhead, we’ll use Redis pub/sub to push core data changes to our main application.

Vert.x also supports direct Redis pub/sub-integration, the following code shows our subscriber-implementation:

vertx.eventBus().<JsonObject>consumer(REDIS_PUBSUB_CHANNEL_VERTX, received -> {

JsonObject value = received.body().getJsonObject("value");

String message = value.getString("message");

JsonObject jsonObject = new JsonObject(message);

eb.send(CACHE_REDIS_EVENTBUS_ADDRESS, jsonObject);

});

redis.subscribe(Constants.REDIS_PUBSUB_CHANNEL, res -> {

if (res.succeeded()) {

LOGGER.info("Subscribed to " + Constants.REDIS_PUBSUB_CHANNEL);

} else {

LOGGER.info(res.cause());

}

});

The verticle subscribes to the appropriate Redis pub/sub-channel. If a message is sent over this channel, the payload is extracted and forwarded to the cache-verticle that stores the data in the L1-cache. After storing and enriching data, a response is sent back to the HttpVerticle, which responds to the HTTP request that initially hit this verticle. In addition, the message is converted to ByteBuffer, wrapped in protocol buffers, and send to an Amazon Kinesis Data Stream.

The following example shows a stripped-down version of the KinesisVerticle:

public class KinesisVerticle extends AbstractVerticle {

private static final Logger LOGGER = LoggerFactory.getLogger(KinesisVerticle.class);

private AmazonKinesisAsync kinesisAsyncClient;

private String eventStream = "EventStream";

@Override

public void start() throws Exception {

EventBus eb = vertx.eventBus();

kinesisAsyncClient = createClient();

eventStream = System.getenv(STREAM_NAME) == null ? "EventStream" : System.getenv(STREAM_NAME);

eb.consumer(Constants.KINESIS_EVENTBUS_ADDRESS, message -> {

try {

TrackingMessage trackingMessage = Json.decodeValue((String)message.body(), TrackingMessage.class);

String partitionKey = trackingMessage.getMessageId();

byte [] byteMessage = createMessage(trackingMessage);

ByteBuffer buf = ByteBuffer.wrap(byteMessage);

sendMessageToKinesis(buf, partitionKey);

message.reply("OK");

}

catch (KinesisException exc) {

LOGGER.error(exc);

}

});

}

Kinesis Consumer
This AWS Lambda function consumes data from an Amazon Kinesis Data Stream and persists the data in an Amazon DynamoDB table. In order to improve testability, the invocation code is separated from the business logic. The invocation code is implemented in the class KinesisConsumerHandler and iterates over the Kinesis events pulled from the Kinesis stream by AWS Lambda. Each Kinesis event is unwrapped and transformed from ByteBuffer to protocol buffers and converted into a Java object. Those Java objects are passed to the business logic, which persists the data in a DynamoDB table. In order to improve duration of successive Lambda calls, the DynamoDB-client is instantiated lazily and reused if possible.

Redis Updater
From time to time, it is necessary to update core data in Redis. A very efficient implementation for this requirement is using AWS Lambda and Amazon Kinesis. New core data is sent over the AWS Kinesis stream using JSON as data format and consumed by a Lambda function. This function iterates over the Kinesis events pulled from the Kinesis stream by AWS Lambda. Each Kinesis event is unwrapped and transformed from ByteBuffer to String and converted into a Java object. The Java object is passed to the business logic and stored in Redis. In addition, the new core data is also sent to the main application using Redis pub/sub in order to reduce network overhead and converting from a pull- to a push-based model.

The following example shows the source code to store data in Redis and notify all subscribers:

public void updateRedisData(final TrackingMessage trackingMessage, final Jedis jedis, final LambdaLogger logger) {

try {

ObjectMapper mapper = new ObjectMapper();

String jsonString = mapper.writeValueAsString(trackingMessage);

Map<String, String> map = marshal(jsonString);

String statusCode = jedis.hmset(trackingMessage.getProgramId(), map);

}

catch (Exception exc) {

if (null == logger)

exc.printStackTrace();

else

logger.log(exc.getMessage());

}

}

public void notifySubscribers(final TrackingMessage trackingMessage, final Jedis jedis, final LambdaLogger logger) {

try {

ObjectMapper mapper = new ObjectMapper();

String jsonString = mapper.writeValueAsString(trackingMessage);

jedis.publish(Constants.REDIS_PUBSUB_CHANNEL, jsonString);

}

catch (final IOException e) {

log(e.getMessage(), logger);

}

}

Similarly to our Kinesis Consumer, the Redis-client is instantiated somewhat lazily.

Infrastructure as Code
As already outlined, latency and response time are a very critical part of any ad-tracking solution because response time has a huge impact on conversion rate. In order to reduce latency for customers world-wide, it is common practice to roll out the infrastructure in different AWS Regions in the world to be as close to the end customer as possible. AWS CloudFormation can help you model and set up your AWS resources so that you can spend less time managing those resources and more time focusing on your applications that run in AWS.

You create a template that describes all the AWS resources that you want (for example, Amazon EC2 instances or Amazon RDS DB instances), and AWS CloudFormation takes care of provisioning and configuring those resources for you. Our reference architecture can be rolled out in different Regions using an AWS CloudFormation template, which sets up the complete infrastructure (for example, Amazon Virtual Private Cloud (Amazon VPC), Amazon Elastic Container Service (Amazon ECS) cluster, Lambda functions, DynamoDB table, Amazon ElastiCache cluster, etc.).

Conclusion
In this blog post we described reactive principles and an example architecture with a common use case. We leveraged the capabilities of different frameworks in combination with several AWS services in order to implement reactive principles—not only at the application-level but also at the system-level. I hope I’ve given you ideas for creating your own reactive applications and systems on AWS.

About the Author

Sascha Moellering is a Senior Solution Architect. Sascha is primarily interested in automation, infrastructure as code, distributed computing, containers and JVM. He can be reached at [email protected]

 

 

Graphite 1.1: Teaching an Old Dog New Tricks

Post Syndicated from Blogs on Grafana Labs Blog original https://grafana.com/blog/2018/01/11/graphite-1.1-teaching-an-old-dog-new-tricks/

The Road to Graphite 1.1

I started working on Graphite just over a year ago, when @obfuscurity asked me to help out with some issues blocking the Graphite 1.0 release. Little did I know that a year later, that would have resulted in 262 commits (and counting), and that with the help of the other Graphite maintainers (especially @deniszh, @iksaif & @cbowman0) we would have added a huge amount of new functionality to Graphite.

There are a huge number of new additions and updates in this release, in this post I’ll give a tour of some of the highlights including tag support, syntax and function updates, custom function plugins, and python 3.x support.

Tagging!

The single biggest feature in this release is the addition of tag support, which brings the ability to describe metrics in a much richer way and to write more flexible and expressive queries.

Traditionally series in Graphite are identified using a hierarchical naming scheme based on dot-separated segments called nodes. This works very well and is simple to map into a hierarchical structure like the whisper filesystem tree, but it means that the user has to know what each segment represents, and makes it very difficult to modify or extend the naming scheme since everything is based on the positions of the segments within the hierarchy.

The tagging system gives users the ability to encode information about the series in a collection of tag=value pairs which are used together with the series name to uniquely identify each series, and the ability to query series by specifying tag-based matching expressions rather than constructing glob-style selectors based on the positions of specific segments within the hierarchy. This is broadly similar to the system used by Prometheus and makes it possible to use Graphite as a long-term storage backend for metrics gathered by Prometheus with full tag support.

When using tags, series names are specified using the new tagged carbon format: name;tag1=value1;tag2=value2. This format is backward compatible with most existing carbon tooling, and makes it easy to adapt existing tools to produce tagged metrics simply by changing the metric names. The OpenMetrics format is also supported for ingestion, and is normalized into the standard Graphite format internally.

At its core, the tagging system is implemented as a tag database (TagDB) alongside the metrics that allows them to be efficiently queried by individual tag values rather than having to traverse the metrics tree looking for series that match the specified query. Internally the tag index is stored in one of a number of pluggable tag databases, currently supported options are the internal graphite-web database, redis, or an external system that implements the Graphite tagging HTTP API. Carbon automatically keeps the index up to date with any tagged series seen.

The new seriesByTag function is used to query the TagDB and will return a list of all the series that match the expressions passed to it. seriesByTag supports both exact and regular expression matches, and can be used anywhere you would previously have specified a metric name or glob expression.

There are new dedicated functions for grouping and aliasing series by tag (groupByTags and aliasByTags), and you can also use tags interchangeably with node numbers in the standard Graphite functions like aliasByNode, groupByNodes, asPercent, mapSeries, etc.

Piping Syntax & Function Updates

One of the huge strengths of the Graphite render API is the ability to chain together multiple functions to process data, but until now (unless you were using a tool like Grafana) writing chained queries could be painful as each function had to be wrapped around the previous one. With this release it is now possible to “pipe” the output of one processing function into the next, and to combine piped and nested functions.

For example:

alias(movingAverage(scaleToSeconds(sumSeries(stats_global.production.counters.api.requests.*.count),60),30),'api.avg')

Can now be written as:

sumSeries(stats_global.production.counters.api.requests.*.count)|scaleToSeconds(60)|movingAverage(30)|alias('api.avg')

OR

stats_global.production.counters.api.requests.*.count|sumSeries()|scaleToSeconds(60)|movingAverage(30)|alias('api.avg')

Another source of frustration with the old function API was the inconsistent implementation of aggregations, with different functions being used in different parts of the API, and some functions simply not being available. In 1.1 all functions that perform aggregation (whether across series or across time intervals) now support a consistent set of aggregations; average, median, sum, min, max, diff, stddev, count, range, multiply and last. This is part of a new approach to implementing functions that emphasises using shared building blocks to ensure consistency across the API and solve the problem of a particular function not working with the aggregation needed for a given task.

To that end a number of new functions have been added that each provide the same functionality as an entire family of “old” functions; aggregate, aggregateWithWildcards, movingWindow, filterSeries, highest, lowest and sortBy.

Each of these functions accepts an aggregation method parameter, for example aggregate(some.metric.*, 'sum') implements the same functionality as sumSeries(some.metric.*).

It can also be used with different aggregation methods to replace averageSeries, stddevSeries, multiplySeries, diffSeries, rangeOfSeries, minSeries, maxSeries and countSeries. All those functions are now implemented as aliases for aggregate, and it supports the previously-missing median and last aggregations.

The same is true for the other functions, and the summarize, smartSummarize, groupByNode, groupByNodes and the new groupByTags functions now all support the standard set of aggregations. Gone are the days of wishing that sortByMedian or highestRange were available!

For more information on the functions available check the function documentation.

Custom Functions

No matter how many functions are available there are always going to be specific use-cases where a custom function can perform analysis that wouldn’t otherwise be possible, or provide a convenient alias for a complicated function chain or specific set of parameters.

In Graphite 1.1 we added support for easily adding one-off custom functions, as well as for creating and sharing plugins that can provide one or more functions.

Each function plugin is packaged as a simple python module, and will be automatically loaded by Graphite when placed into the functions/custom folder.

An example of a simple function plugin that translates the name of every series passed to it into UPPERCASE:

from graphite.functions.params import Param, ParamTypes

def toUpperCase(requestContext, seriesList):
  """Custom function that changes series names to UPPERCASE"""
  for series in seriesList:
    series.name = series.name.upper()
  return seriesList

toUpperCase.group = 'Custom'
toUpperCase.params = [
  Param('seriesList', ParamTypes.seriesList, required=True),
]

SeriesFunctions = {
  'upper': toUpperCase,
}

Once installed the function is not only available for use within Grpahite, but is also exposed via the new Function API which allows the function definition and documentation to be automatically loaded by tools like Grafana. This means that users will be able to select and use the new function in exactly the same way as the internal functions.

More information on writing and using custom functions is available in the documentation.

Clustering Updates

One of the biggest changes from the 0.9 to 1.0 releases was the overhaul of the clustering code, and with 1.1.1 that process has been taken even further to optimize performance when using Graphite in a clustered deployment. In the past it was common for a request to require the frontend node to make multiple requests to the backend nodes to identify matching series and to fetch data, and the code for handling remote vs local series was overly complicated. In 1.1.1 we took a new approach where all render data requests pass through the same path internally, and multiple backend nodes are handled individually rather than grouped together into a single finder. This has greatly simplified the codebase, making it much easier to understand and reason about, while allowing much more flexibility in design of the finders. After these changes, render requests can now be answered with a single internal request to each backend node, and all requests for both remote and local data are executed in parallel.

To maintain the ability of graphite to scale out horizontally, the tagging system works seamlessly within a clustered environment, with each node responsible for the series stored on that node. Calls to load tagged series via seriesByTag are fanned out to the backend nodes and results are merged on the query node just like they are for non-tagged series.

Python 3 & Django 1.11 Support

Graphite 1.1 finally brings support for Python 3.x, both graphite-web and carbon are now tested against Python 2.7, 3.4, 3.5, 3.6 and PyPy. Django releases 1.8 through 1.11 are also supported. The work involved in sorting out the compatibility issues between Python 2.x and 3.x was quite involved, but it is a huge step forward for the long term support of the project! With the new Django 2.x series supporting only Python 3.x we will need to evaluate our long-term support for Python 2.x, but the Django 1.11 series is supported through 2020 so there is time to consider the options there.

Watch This Space

Efforts are underway to add support for the new functionality across the ecosystem of tools that work with Graphite, adding collectd tagging support, prometheus remote read & write with tags (and native Prometheus remote read/write support in Graphite) and last but not least Graphite tag support in Grafana.

We’re excited about the possibilities that the new capabilities in 1.1.x open up, and can’t wait to see how the community puts them to work.

Download the 1.1.1 release and check out the release notes here.

[$] A new kernel polling interface

Post Syndicated from corbet original https://lwn.net/Articles/743714/rss

Polling a set of file descriptors to see which ones can perform I/O without
blocking is a useful thing to do — so useful that the kernel provides three
different system calls (select(),
poll(),
and epoll_wait()
— plus some variants) to perform it. But sometimes three is not enough;
there is now a proposal circulating for a fourth kernel polling interface.
As is usually the case, the motivation for this change is performance.

Spectre and Meltdown Attacks Against Microprocessors

Post Syndicated from Bruce Schneier original https://www.schneier.com/blog/archives/2018/01/spectre_and_mel_1.html

The security of pretty much every computer on the planet has just gotten a lot worse, and the only real solution — which of course is not a solution — is to throw them all away and buy new ones.

On Wednesday, researchers just announced a series of major security vulnerabilities in the microprocessors at the heart of the world’s computers for the past 15-20 years. They’ve been named Spectre and Meltdown, and they have to do with manipulating different ways processors optimize performance by rearranging the order of instructions or performing different instructions in parallel. An attacker who controls one process on a system can use the vulnerabilities to steal secrets elsewhere on the computer. (The research papers are here and here.)

This means that a malicious app on your phone could steal data from your other apps. Or a malicious program on your computer — maybe one running in a browser window from that sketchy site you’re visiting, or as a result of a phishing attack — can steal data elsewhere on your machine. Cloud services, which often share machines amongst several customers, are especially vulnerable. This affects corporate applications running on cloud infrastructure, and end-user cloud applications like Google Drive. Someone can run a process in the cloud and steal data from every other users on the same hardware.

Information about these flaws has been secretly circulating amongst the major IT companies for months as they researched the ramifications and coordinated updates. The details were supposed to be released next week, but the story broke early and everyone is scrambling. By now all the major cloud vendors have patched their systems against the vulnerabilities that can be patched against.

“Throw it away and buy a new one” is ridiculous security advice, but it’s what US-CERT recommends. It is also unworkable. The problem is that there isn’t anything to buy that isn’t vulnerable. Pretty much every major processor made in the past 20 years is vulnerable to some flavor of these vulnerabilities. Patching against Meltdown can degrade performance by almost a third. And there’s no patch for Spectre; the microprocessors have to be redesigned to prevent the attack, and that will take years. (Here’s a running list of who’s patched what.)

This is bad, but expect it more and more. Several trends are converging in a way that makes our current system of patching security vulnerabilities harder to implement.

The first is that these vulnerabilities affect embedded computers in consumer devices. Unlike our computer and phones, these systems are designed and produced at a lower profit margin with less engineering expertise. There aren’t security teams on call to write patches, and there often aren’t mechanisms to push patches onto the devices. We’re already seeing this with home routers, digital video recorders, and webcams. The vulnerability that allowed them to be taken over by the Mirai botnet last August simply can’t be fixed.

The second is that some of the patches require updating the computer’s firmware. This is much harder to walk consumers through, and is more likely to permanently brick the device if something goes wrong. It also requires more coordination. In November, Intel released a firmware update to fix a vulnerability in its Management Engine (ME): another flaw in its microprocessors. But it couldn’t get that update directly to users; it had to work with the individual hardware companies, and some of them just weren’t capable of getting the update to their customers.

We’re already seeing this. Some patches require users to disable the computer’s password, which means organizations can’t automate the patch. Some antivirus software blocks the patch, or — worse — crashes the computer. This results in a three-step process: patch your antivirus software, patch your operating system, and then patch the computer’s firmware.

The final reason is the nature of these vulnerabilities themselves. These aren’t normal software vulnerabilities, where a patch fixes the problem and everyone can move on. These vulnerabilities are in the fundamentals of how the microprocessor operates.

It shouldn’t be surprising that microprocessor designers have been building insecure hardware for 20 years. What’s surprising is that it took 20 years to discover it. In their rush to make computers faster, they weren’t thinking about security. They didn’t have the expertise to find these vulnerabilities. And those who did were too busy finding normal software vulnerabilities to examine microprocessors. Security researchers are starting to look more closely at these systems, so expect to hear about more vulnerabilities along these lines.

Spectre and Meltdown are pretty catastrophic vulnerabilities, but they only affect the confidentiality of data. Now that they — and the research into the Intel ME vulnerability — have shown researchers where to look, more is coming — and what they’ll find will be worse than either Spectre or Meltdown. There will be vulnerabilities that will allow attackers to manipulate or delete data across processes, potentially fatal in the computers controlling our cars or implanted medical devices. These will be similarly impossible to fix, and the only strategy will be to throw our devices away and buy new ones.

This isn’t to say you should immediately turn your computers and phones off and not use them for a few years. For the average user, this is just another attack method amongst many. All the major vendors are working on patches and workarounds for the attacks they can mitigate. All the normal security advice still applies: watch for phishing attacks, don’t click on strange e-mail attachments, don’t visit sketchy websites that might run malware on your browser, patch your systems regularly, and generally be careful on the Internet.

You probably won’t notice that performance hit once Meltdown is patched, except maybe in backup programs and networking applications. Embedded systems that do only one task, like your programmable thermostat or the computer in your refrigerator, are unaffected. Small microprocessors that don’t do all of the vulnerable fancy performance tricks are unaffected. Browsers will figure out how to mitigate this in software. Overall, the security of the average Internet-of-Things device is so bad that this attack is in the noise compared to the previously known risks.

It’s a much bigger problem for cloud vendors; the performance hit will be expensive, but I expect that they’ll figure out some clever way of detecting and blocking the attacks. All in all, as bad as Spectre and Meltdown are, I think we got lucky.

But more are coming, and they’ll be worse. 2018 will be the year of microprocessor vulnerabilities, and it’s going to be a wild ride.

Note: A shorter version of this essay previously appeared on CNN.com. My previous blog post on this topic contains additional links.

Detecting Adblocker Blockers

Post Syndicated from Bruce Schneier original https://www.schneier.com/blog/archives/2018/01/detecting_adblo.html

Interesting research on the prevalence of adblock blockers: “Measuring and Disrupting Anti-Adblockers Using Differential Execution Analysis“:

Abstract: Millions of people use adblockers to remove intrusive and malicious ads as well as protect themselves against tracking and pervasive surveillance. Online publishers consider adblockers a major threat to the ad-powered “free” Web. They have started to retaliate against adblockers by employing anti-adblockers which can detect and stop adblock users. To counter this retaliation, adblockers in turn try to detect and filter anti-adblocking scripts. This back and forth has prompted an escalating arms race between adblockers and anti-adblockers.

We want to develop a comprehensive understanding of anti-adblockers, with the ultimate aim of enabling adblockers to bypass state-of-the-art anti-adblockers. In this paper, we present a differential execution analysis to automatically detect and analyze anti-adblockers. At a high level, we collect execution traces by visiting a website with and without adblockers. Through differential execution analysis, we are able to pinpoint the conditions that lead to the differences caused by anti-adblocking code. Using our system, we detect anti-adblockers on 30.5% of the Alexa top-10K websites which is 5-52 times more than reported in prior literature. Unlike prior work which is limited to detecting visible reactions (e.g., warning messages) by anti-adblockers, our system can discover attempts to detect adblockers even when there is no visible reaction. From manually checking one third of the detected websites, we find that the websites that have no visible reactions constitute over 90% of the cases, completely dominating the ones that have visible warning messages. Finally, based on our findings, we further develop JavaScript rewriting and API hooking based solutions (the latter implemented as a Chrome extension) to help adblockers bypass state-of-the-art anti-adblockers.

News article.

Goodbye, net neutrality—Ajit Pai’s FCC votes to allow blocking and throttling (Ars Technica)

Post Syndicated from jake original https://lwn.net/Articles/741482/rss

In a vote that was not any kind of surprise, the US Federal Communications Commission (FCC) voted to end the “net neutrality” rules that stop internet service providers (ISPs) and others from blocking or throttling certain kinds of traffic to try to force consumers and content providers to pay more for “fast lanes”. Ars Technica covers the vote and the reaction to it, including the fact that the fight is not yet over: “Plenty of organizations might appeal, said consumer advocate Gigi Sohn, who was a top counselor to then-FCC Chairman Tom Wheeler when the commission imposed its rules.

‘I think you’ll see public interest groups, trade associations, and small and mid-sized tech companies filing the petitions for review,’ Sohn told Ars. One or two ‘big companies’ could also challenge the repeal, she thinks.

Lawsuit filers can challenge the repeal on numerous respects, she said. They can argue that the public record doesn’t support the FCC’s claim that broadband isn’t a telecommunications service, that ‘throwing away all protections for consumers and innovators for the first time since this issue has been debated is arbitrary and capricious,’ and that the FCC cannot preempt state net neutrality laws, she said.”

Libertarians are against net neutrality

Post Syndicated from Robert Graham original http://blog.erratasec.com/2017/12/libertarians-are-against-net-neutrality.html

This post claims to be by a libertarian in support of net neutrality. As a libertarian, I need to debunk this. “Net neutrality” is a case of one-hand clapping, you rarely hear the competing side, and thus, that side may sound attractive. This post is about the other side, from a libertarian point of view.

That post just repeats the common, and wrong, left-wing talking points. I mean, there might be a libertarian case for some broadband regulation, but this isn’t it.

This thing they call “net neutrality” is just left-wing politics masquerading as some sort of principle. It’s no different than how people claim to be “pro-choice”, yet demand forced vaccinations. Or, it’s no different than how people claim to believe in “traditional marriage” even while they are on their third “traditional marriage”.

Properly defined, “net neutrality” means no discrimination of network traffic. But nobody wants that. A classic example is how most internet connections have faster download speeds than uploads. This discriminates against upload traffic, harming innovation in upload-centric applications like DropBox’s cloud backup or BitTorrent’s peer-to-peer file transfer. Yet activists never mention this, or other types of network traffic discrimination, because they no more care about “net neutrality” than Trump or Gingrich care about “traditional marriage”.

Instead, when people say “net neutrality”, they mean “government regulation”. It’s the same old debate between who is the best steward of consumer interest: the free-market or government.

Specifically, in the current debate, they are referring to the Obama-era FCC “Open Internet” order and reclassification of broadband under “Title II” so they can regulate it. Trump’s FCC is putting broadband back to “Title I”, which means the FCC can’t regulate most of its “Open Internet” order.

Don’t be tricked into thinking the “Open Internet” order is anything but intensely politically. The premise behind the order is the Democrat’s firm believe that it’s government who created the Internet, and all innovation, advances, and investment ultimately come from the government. It sees ISPs as inherently deceitful entities who will only serve their own interests, at the expense of consumers, unless the FCC protects consumers.

It says so right in the order itself. It starts with the premise that broadband ISPs are evil, using illegitimate “tactics” to hurt consumers, and continues with similar language throughout the order.

A good contrast to this can be seen in Tim Wu’s non-political original paper in 2003 that coined the term “net neutrality”. Whereas the FCC sees broadband ISPs as enemies of consumers, Wu saw them as allies. His concern was not that ISPs would do evil things, but that they would do stupid things, such as favoring short-term interests over long-term innovation (such as having faster downloads than uploads).

The political depravity of the FCC’s order can be seen in this comment from one of the commissioners who voted for those rules:

FCC Commissioner Jessica Rosenworcel wants to increase the minimum broadband standards far past the new 25Mbps download threshold, up to 100Mbps. “We invented the internet. We can do audacious things if we set big goals, and I think our new threshold, frankly, should be 100Mbps. I think anything short of that shortchanges our children, our future, and our new digital economy,” Commissioner Rosenworcel said.

This is indistinguishable from communist rhetoric that credits the Party for everything, as this booklet from North Korea will explain to you.

But what about monopolies? After all, while the free-market may work when there’s competition, it breaks down where there are fewer competitors, oligopolies, and monopolies.

There is some truth to this, in individual cities, there’s often only only a single credible high-speed broadband provider. But this isn’t the issue at stake here. The FCC isn’t proposing light-handed regulation to keep monopolies in check, but heavy-handed regulation that regulates every last decision.

Advocates of FCC regulation keep pointing how broadband monopolies can exploit their renting-seeking positions in order to screw the customer. They keep coming up with ever more bizarre and unlikely scenarios what monopoly power grants the ISPs.

But the never mention the most simplest: that broadband monopolies can just charge customers more money. They imagine instead that these companies will pursue a string of outrageous, evil, and less profitable behaviors to exploit their monopoly position.

The FCC’s reclassification of broadband under Title II gives it full power to regulate ISPs as utilities, including setting prices. The FCC has stepped back from this, promising it won’t go so far as to set prices, that it’s only regulating these evil conspiracy theories. This is kind of bizarre: either broadband ISPs are evilly exploiting their monopoly power or they aren’t. Why stop at regulating only half the evil?

The answer is that the claim “monopoly” power is a deception. It starts with overstating how many monopolies there are to begin with. When it issued its 2015 “Open Internet” order the FCC simultaneously redefined what they meant by “broadband”, upping the speed from 5-mbps to 25-mbps. That’s because while most consumers have multiple choices at 5-mbps, fewer consumers have multiple choices at 25-mbps. It’s a dirty political trick to convince you there is more of a problem than there is.

In any case, their rules still apply to the slower broadband providers, and equally apply to the mobile (cell phone) providers. The US has four mobile phone providers (AT&T, Verizon, T-Mobile, and Sprint) and plenty of competition between them. That it’s monopolistic power that the FCC cares about here is a lie. As their Open Internet order clearly shows, the fundamental principle that animates the document is that all corporations, monopolies or not, are treacherous and must be regulated.

“But corporations are indeed evil”, people argue, “see here’s a list of evil things they have done in the past!”

No, those things weren’t evil. They were done because they benefited the customers, not as some sort of secret rent seeking behavior.

For example, one of the more common “net neutrality abuses” that people mention is AT&T’s blocking of FaceTime. I’ve debunked this elsewhere on this blog, but the summary is this: there was no network blocking involved (not a “net neutrality” issue), and the FCC analyzed it and decided it was in the best interests of the consumer. It’s disingenuous to claim it’s an evil that justifies FCC actions when the FCC itself declared it not evil and took no action. It’s disingenuous to cite the “net neutrality” principle that all network traffic must be treated when, in fact, the network did treat all the traffic equally.

Another frequently cited abuse is Comcast’s throttling of BitTorrent.Comcast did this because Netflix users were complaining. Like all streaming video, Netflix backs off to slower speed (and poorer quality) when it experiences congestion. BitTorrent, uniquely among applications, never backs off. As most applications become slower and slower, BitTorrent just speeds up, consuming all available bandwidth. This is especially problematic when there’s limited upload bandwidth available. Thus, Comcast throttled BitTorrent during prime time TV viewing hours when the network was already overloaded by Netflix and other streams. BitTorrent users wouldn’t mind this throttling, because it often took days to download a big file anyway.

When the FCC took action, Comcast stopped the throttling and imposed bandwidth caps instead. This was a worse solution for everyone. It penalized heavy Netflix viewers, and prevented BitTorrent users from large downloads. Even though BitTorrent users were seen as the victims of this throttling, they’d vastly prefer the throttling over the bandwidth caps.

In both the FaceTime and BitTorrent cases, the issue was “network management”. AT&T had no competing video calling service, Comcast had no competing download service. They were only reacting to the fact their networks were overloaded, and did appropriate things to solve the problem.

Mobile carriers still struggle with the “network management” issue. While their networks are fast, they are still of low capacity, and quickly degrade under heavy use. They are looking for tricks in order to reduce usage while giving consumers maximum utility.

The biggest concern is video. It’s problematic because it’s designed to consume as much bandwidth as it can, throttling itself only when it experiences congestion. This is what you probably want when watching Netflix at the highest possible quality, but it’s bad when confronted with mobile bandwidth caps.

With small mobile devices, you don’t want as much quality anyway. You want the video degraded to lower quality, and lower bandwidth, all the time.

That’s the reasoning behind T-Mobile’s offerings. They offer an unlimited video plan in conjunction with the biggest video providers (Netflix, YouTube, etc.). The catch is that when congestion occurs, they’ll throttle it to lower quality. In other words, they give their bandwidth to all the other phones in your area first, then give you as much of the leftover bandwidth as you want for video.

While it sounds like T-Mobile is doing something evil, “zero-rating” certain video providers and degrading video quality, the FCC allows this, because they recognize it’s in the customer interest.

Mobile providers especially have great interest in more innovation in this area, in order to conserve precious bandwidth, but they are finding it costly. They can’t just innovate, but must ask the FCC permission first. And with the new heavy handed FCC rules, they’ve become hostile to this innovation. This attitude is highlighted by the statement from the “Open Internet” order:

And consumers must be protected, for example from mobile commercial practices masquerading as “reasonable network management.”

This is a clear declaration that free-market doesn’t work and won’t correct abuses, and that that mobile companies are treacherous and will do evil things without FCC oversight.

Conclusion

Ignoring the rhetoric for the moment, the debate comes down to simple left-wing authoritarianism and libertarian principles. The Obama administration created a regulatory regime under clear Democrat principles, and the Trump administration is rolling it back to more free-market principles. There is no principle at stake here, certainly nothing to do with a technical definition of “net neutrality”.

The 2015 “Open Internet” order is not about “treating network traffic neutrally”, because it doesn’t do that. Instead, it’s purely a left-wing document that claims corporations cannot be trusted, must be regulated, and that innovation and prosperity comes from the regulators and not the free market.

It’s not about monopolistic power. The primary targets of regulation are the mobile broadband providers, where there is plenty of competition, and who have the most “network management” issues. Even if it were just about wired broadband (like Comcast), it’s still ignoring the primary ways monopolies profit (raising prices) and instead focuses on bizarre and unlikely ways of rent seeking.

If you are a libertarian who nonetheless believes in this “net neutrality” slogan, you’ve got to do better than mindlessly repeating the arguments of the left-wing. The term itself, “net neutrality”, is just a slogan, varying from person to person, from moment to moment. You have to be more specific. If you truly believe in the “net neutrality” technical principle that all traffic should be treated equally, then you’ll want a rewrite of the “Open Internet” order.

In the end, while libertarians may still support some form of broadband regulation, it’s impossible to reconcile libertarianism with the 2015 “Open Internet”, or the vague things people mean by the slogan “net neutrality”.

AWS Media Services – Process, Store, and Monetize Cloud-Based Video

Post Syndicated from Jeff Barr original https://aws.amazon.com/blogs/aws/aws-media-services-process-store-and-monetize-cloud-based-video/

Do you remember what web video was like in the early days? Standalone players, video no larger than a postage stamp, slow & cantankerous connections, overloaded servers, and the ever-present buffering messages were the norm less than two decades ago.

Today, thanks to technological progress and a broad array of standards, things are a lot better. Video consumers are now in control. They use devices of all shapes, sizes, and vintages to enjoy live and recorded content that is broadcast, streamed, or sent over-the-top (OTT, as they say), and expect immediate access to content that captures and then holds their attention. Meeting these expectations presents a challenge for content creators and distributors. Instead of generating video in a one-size-fits-all format, they (or their media servers) must be prepared to produce video that spans a broad range of sizes, formats, and bit rates, taking care to be ready to deal with planned or unplanned surges in demand. In the face of all of this complexity, they must backstop their content with a monetization model that supports the content and the infrastructure to deliver it.

New AWS Media Services
Today we are launching an array of broadcast-quality media services, each designed to address one or more aspects of the challenge that I outlined above. You can use them together to build a complete end-to-end video solution or you can use one or more in building-block style. In true AWS fashion, you can spend more time innovating and less time setting up and running infrastructure, leaving you ready to focus on creating, delivering, and monetizing your content. The services are all elastic, allowing you to ramp up processing power, connections, and storage and giving you the ability to handle million-user (and beyond) spikes with ease.

Here are the services (all accessible from a set of interactive consoles as well as through a comprehensive set of APIs):

AWS Elemental MediaConvert – File-based transcoding for OTT, broadcast, or archiving, with support for a long list of formats and codecs. Features include multi-channel audio, graphic overlays, closed captioning, and several DRM options.

AWS Elemental MediaLive – Live encoding to deliver video streams in real time to both televisions and multiscreen devices. Allows you to deploy highly reliable live channels in minutes, with full control over encoding parameters. It supports ad insertion, multi-channel audio, graphic overlays, and closed captioning.

AWS Elemental MediaPackage – Video origination and just-in-time packaging. Starting from a single input, produces output for multiple devices representing a long list of current and legacy formats. Supports multiple monetization models, time-shifted live streaming, ad insertion, DRM, and blackout management.

AWS Elemental MediaStore – Media-optimized storage that enables high performance and low latency applications such as live streaming, while taking advantage of the scale and durability of Amazon Simple Storage Service (S3).

AWS Elemental MediaTailor – Monetization service that supports ad serving and server-side ad insertion, a broad range of devices, transcoding, and accurate reporting of server-side and client-side ad insertion.

Instead of listing out all of the features in the sections below, I’ve simply included as many screen shots as possible with the expectation that this will give you a better sense of the rich set of features, parameters, and settings that you get with this set of services.

AWS Elemental MediaConvert
MediaConvert allows you to transcode content that is stored in files. You can process individual files or entire media libraries, or anything in-between. You simply create a conversion job that specifies the content and the desired outputs, and submit it to MediaConvert. There’s no software to install or patch and the service scales to meet your needs without affecting turnaround time or performance.

The MediaConvert Console lets you manage Output presets, Job templates, Queues, and Jobs:

You can use a built-in system preset or you can make one of your own. You have full control over the settings when you make your own:

Jobs templates are named, and produce one or more output groups. You can add a new group to a template with a click:

When everything is ready to go, you create a job and make some final selections, then click on Create:

Each account starts with a default queue for jobs, where incoming work is processed in parallel using all processing resources available to the account. Adding queues does not add processing resources, but does cause them to be apportioned across queues. You can temporarily pause one queue in order to devote more resources to the others. You can submit jobs to paused queues and you can also cancel any that have yet to start.

Pricing for this service is based on the amount of video that you process and the features that you use.

AWS Elemental MediaLive
This service is for live encoding, and can be run 24×7. MediaLive channels are deployed on redundant resources distributed in two physically separated Availability Zones in order to provide the reliability expected by our customers in the broadcast industry. You can specify your inputs and define your channels in the MediaLive Console:

After you create an Input, you create a Channel and attach it to the Input:

You have full control over the settings for each channel:

 

AWS Elemental MediaPackage
This service lets you deliver video to many devices from a single source. It focuses on protection and just-in-time packaging, giving you the ability to provide your users with the desired content on the device of their choice. You simply create a channel to get started:

Then you add one or more endpoints. Once again, plenty of options and full control, including a startover window and a time delay:

You find the input URL, user name, and password for your channel and route your live video stream to it for packaging:

AWS Elemental MediaStore
MediaStore offers the performance, consistency, and latency required for live and on-demand media delivery. Objects are written and read into a new “temporal” tier of object storage for a limited amount of time, then move silently into S3 for long-lived durability. You simply create a storage container to group your media content:

The container is available within a minute or so:

Like S3 buckets, MediaStore containers have access policies and no limits on the number of objects or storage capacity.

MediaStore helps you to take full advantage of S3 by managing the object key names so as to maximize storage and retrieval throughput, in accord with the Request Rate and Performance Considerations.

AWS Elemental MediaTailor
This service takes care of server-side ad insertion while providing a broadcast-quality viewer experience by transcoding ad assets on the fly. Your customer’s video player asks MediaTailor for a playlist. MediaTailor, in turn, calls your Ad Decision Server and returns a playlist that references the origin server for your original video and the ads recommended by the Ad Decision Server. The video player makes all of its requests to a single endpoint in order to ensure that client-side ad-blocking is ineffective. You simply create a MediaTailor Configuration:

Context information is passed to the Ad Decision Server in the URL:

Despite the length of this post I have barely scratched the surface of the AWS Media Services. Once AWS re:Invent is in the rear view mirror I hope to do a deep dive and show you how to use each of these services.

Available Now
The entire set of AWS Media Services is available now and you can start using them today! Pricing varies by service, but is built around a pay-as-you-go model.

Jeff;

NetNeutrality vs. limiting FaceTime

Post Syndicated from Robert Graham original http://blog.erratasec.com/2017/11/netneutrality-vs-limiting-facetime.html

People keep retweeting this ACLU graphic in regards to NetNeutrality. In this post, I debunk the fourth item. In previous posts [1] [2] I debunk other items.

But here’s the thing: the FCC allowed these restrictions, despite the FCC’s “Open Internet” order forbidding such things. In other words, despite the graphic’s claims it “happened without net neutrality rules”, the opposite is true, it happened with net neutrality rules.

The FCC explains why they allowed it in their own case study on the matter. The short version is this: AT&T’s network couldn’t handle the traffic, so it was appropriate to restrict it until some time in the future (the LTE rollout) until it could. The issue wasn’t that AT&T was restricting FaceTime in favor of its own video-calling service (it didn’t have one), but it was instead an issue of “bandwidth management”.
When Apple released FaceTime, they themselves restricted it’s use to WiFi, preventing its use on cell phone networks. That’s because Apple recognized mobile networks couldn’t handle it.
When Apple flipped the switch and allowed it’s use on mobile networks, because mobile networks had gotten faster, they clearly said “carrier restrictions may apply”. In other words, it said “carriers may restrict FaceTime with our blessing if they can’t handle the load”.
When Tim Wu wrote his paper defining “NetNeutrality” in 2003, he anticipated just this scenario. He wrote:

“The goal of bandwidth management is, at a general level, aligned with network neutrality.”

He doesn’t give “bandwidth management” a completely free pass. He mentions the issue frequently in his paper with a less favorable description, such as here:

Similarly, while managing bandwidth is a laudable goal, its achievement through restricting certain application types is an unfortunate solution. The result is obviously a selective disadvantage for certain application markets. The less restrictive means is, as above, the technological management of bandwidth. Application-restrictions should, at best, be a stopgap solution to the problem of competing bandwidth demands. 

And that’s what AT&T’s FaceTime limiting was: an unfortunate stopgap solution until LTE was more fully deployed, which is fully allowed under Tim Wu’s principle of NetNeutrality.

So the ACLU’s claim above is fully debunked: such things did happen even with NetNeutrality rules in place, and should happen.

Finally, and this is probably the most important part, AT&T didn’t block it in the network. Instead, they blocked the app on the phone. If you jailbroke your phone, you could use FaceTime as you wished. Thus, it’s not a “network” neutrality issue because no blocking happened in the network.

Introducing Cloud Native Networking for Amazon ECS Containers

Post Syndicated from Nathan Taber original https://aws.amazon.com/blogs/compute/introducing-cloud-native-networking-for-ecs-containers/

This post courtesy of ECS Sr. Software Dev Engineer Anirudh Aithal.

Today, AWS announced Task Networking for Amazon ECS. This feature brings Amazon EC2 networking capabilities to tasks using elastic network interfaces.

An elastic network interface is a virtual network interface that you can attach to an instance in a VPC. When you launch an EC2 virtual machine, an elastic network interface is automatically provisioned to provide networking capabilities for the instance.

A task is a logical group of running containers. Previously, tasks running on Amazon ECS shared the elastic network interface of their EC2 host. Now, the new awsvpc networking mode lets you attach an elastic network interface directly to a task.

This simplifies network configuration, allowing you to treat each container just like an EC2 instance with full networking features, segmentation, and security controls in the VPC.

In this post, I cover how awsvpc mode works and show you how you can start using elastic network interfaces with your tasks running on ECS.

Background:  Elastic network interfaces in EC2

When you launch EC2 instances within a VPC, you don’t have to configure an additional overlay network for those instances to communicate with each other. By default, routing tables in the VPC enable seamless communication between instances and other endpoints. This is made possible by virtual network interfaces in VPCs called elastic network interfaces. Every EC2 instance that launches is automatically assigned an elastic network interface (the primary network interface). All networking parameters—such as subnets, security groups, and so on—are handled as properties of this primary network interface.

Furthermore, an IPv4 address is allocated to every elastic network interface by the VPC at creation (the primary IPv4 address). This primary address is unique and routable within the VPC. This effectively makes your VPC a flat network, resulting in a simple networking topology.

Elastic network interfaces can be treated as fundamental building blocks for connecting various endpoints in a VPC, upon which you can build higher-level abstractions. This allows elastic network interfaces to be leveraged for:

  • VPC-native IPv4 addressing and routing (between instances and other endpoints in the VPC)
  • Network traffic isolation
  • Network policy enforcement using ACLs and firewall rules (security groups)
  • IPv4 address range enforcement (via subnet CIDRs)

Why use awsvpc?

Previously, ECS relied on the networking capability provided by Docker’s default networking behavior to set up the network stack for containers. With the default bridge network mode, containers on an instance are connected to each other using the docker0 bridge. Containers use this bridge to communicate with endpoints outside of the instance, using the primary elastic network interface of the instance on which they are running. Containers share and rely on the networking properties of the primary elastic network interface, including the firewall rules (security group subscription) and IP addressing.

This means you cannot address these containers with the IP address allocated by Docker (it’s allocated from a pool of locally scoped addresses), nor can you enforce finely grained network ACLs and firewall rules. Instead, containers are addressable in your VPC by the combination of the IP address of the primary elastic network interface of the instance, and the host port to which they are mapped (either via static or dynamic port mapping). Also, because a single elastic network interface is shared by multiple containers, it can be difficult to create easily understandable network policies for each container.

The awsvpc networking mode addresses these issues by provisioning elastic network interfaces on a per-task basis. Hence, containers no longer share or contend use these resources. This enables you to:

  • Run multiple copies of the container on the same instance using the same container port without needing to do any port mapping or translation, simplifying the application architecture.
  • Extract higher network performance from your applications as they no longer contend for bandwidth on a shared bridge.
  • Enforce finer-grained access controls for your containerized applications by associating security group rules for each Amazon ECS task, thus improving the security for your applications.

Associating security group rules with a container or containers in a task allows you to restrict the ports and IP addresses from which your application accepts network traffic. For example, you can enforce a policy allowing SSH access to your instance, but blocking the same for containers. Alternatively, you could also enforce a policy where you allow HTTP traffic on port 80 for your containers, but block the same for your instances. Enforcing such security group rules greatly reduces the surface area of attack for your instances and containers.

ECS manages the lifecycle and provisioning of elastic network interfaces for your tasks, creating them on-demand and cleaning them up after your tasks stop. You can specify the same properties for the task as you would when launching an EC2 instance. This means that containers in such tasks are:

  • Addressable by IP addresses and the DNS name of the elastic network interface
  • Attachable as ‘IP’ targets to Application Load Balancers and Network Load Balancers
  • Observable from VPC flow logs
  • Access controlled by security groups

­This also enables you to run multiple copies of the same task definition on the same instance, without needing to worry about port conflicts. You benefit from higher performance because you don’t need to perform any port translations or contend for bandwidth on the shared docker0 bridge, as you do with the bridge networking mode.

Getting started

If you don’t already have an ECS cluster, you can create one using the create cluster wizard. In this post, I use “awsvpc-demo” as the cluster name. Also, if you are following along with the command line instructions, make sure that you have the latest version of the AWS CLI or SDK.

Registering the task definition

The only change to make in your task definition for task networking is to set the networkMode parameter to awsvpc. In the ECS console, enter this value for Network Mode.

 

If you plan on registering a container in this task definition with an ECS service, also specify a container port in the task definition. This example specifies an NGINX container exposing port 80:

This creates a task definition named “nginx-awsvpc" with networking mode set to awsvpc. The following commands illustrate registering the task definition from the command line:

$ cat nginx-awsvpc.json
{
        "family": "nginx-awsvpc",
        "networkMode": "awsvpc",
        "containerDefinitions": [
            {
                "name": "nginx",
                "image": "nginx:latest",
                "cpu": 100,
                "memory": 512,
                "essential": true,
                "portMappings": [
                  {
                    "containerPort": 80,
                    "protocol": "tcp"
                  }
                ]
            }
        ]
}

$ aws ecs register-task-definition --cli-input-json file://./nginx-awsvpc.json

Running the task

To run a task with this task definition, navigate to the cluster in the Amazon ECS console and choose Run new task. Specify the task definition as “nginx-awsvpc“. Next, specify the set of subnets in which to run this task. You must have instances registered with ECS in at least one of these subnets. Otherwise, ECS can’t find a candidate instance to attach the elastic network interface.

You can use the console to narrow down the subnets by selecting a value for Cluster VPC:

 

Next, select a security group for the task. For the purposes of this example, create a new security group that allows ingress only on port 80. Alternatively, you can also select security groups that you’ve already created.

Next, run the task by choosing Run Task.

You should have a running task now. If you look at the details of the task, you see that it has an elastic network interface allocated to it, along with the IP address of the elastic network interface:

You can also use the command line to do this:

$ aws ecs run-task --cluster awsvpc-ecs-demo --network-configuration "awsvpcConfiguration={subnets=["subnet-c070009b"],securityGroups=["sg-9effe8e4"]}" nginx-awsvpc $ aws ecs describe-tasks --cluster awsvpc-ecs-demo --task $ECS_TASK_ARN --query tasks[0]
{
    "taskArn": "arn:aws:ecs:us-west-2:xx..x:task/f5xx-...",
    "group": "family:nginx-awsvpc",
    "attachments": [
        {
            "status": "ATTACHED",
            "type": "ElasticNetworkInterface",
            "id": "xx..",
            "details": [
                {
                    "name": "subnetId",
                    "value": "subnet-c070009b"
                },
                {
                    "name": "networkInterfaceId",
                    "value": "eni-b0aaa4b2"
                },
                {
                    "name": "macAddress",
                    "value": "0a:47:e4:7a:2b:02"
                },
                {
                    "name": "privateIPv4Address",
                    "value": "10.0.0.35"
                }
            ]
        }
    ],
    ...
    "desiredStatus": "RUNNING",
    "taskDefinitionArn": "arn:aws:ecs:us-west-2:xx..x:task-definition/nginx-awsvpc:2",
    "containers": [
        {
            "containerArn": "arn:aws:ecs:us-west-2:xx..x:container/62xx-...",
            "taskArn": "arn:aws:ecs:us-west-2:xx..x:task/f5x-...",
            "name": "nginx",
            "networkBindings": [],
            "lastStatus": "RUNNING",
            "networkInterfaces": [
                {
                    "privateIpv4Address": "10.0.0.35",
                    "attachmentId": "xx.."
                }
            ]
        }
    ]
}

When you describe an “awsvpc” task, details of the elastic network interface are returned via the “attachments” object. You can also get this information from the “containers” object. For example:

$ aws ecs describe-tasks --cluster awsvpc-ecs-demo --task $ECS_TASK_ARN --query tasks[0].containers[0].networkInterfaces[0].privateIpv4Address
"10.0.0.35"

Conclusion

The nginx container is now addressable in your VPC via the 10.0.0.35 IPv4 address. You did not have to modify the security group on the instance to allow requests on port 80, thus improving instance security. Also, you ensured that all ports apart from port 80 were blocked for this application without modifying the application itself, which makes it easier to manage your task on the network. You did not have to interact with any of the elastic network interface API operations, as ECS handled all of that for you.

You can read more about the task networking feature in the ECS documentation. For a detailed look at how this new networking mode is implemented on an instance, see Under the Hood: Task Networking for Amazon ECS.

Please use the comments section below to send your feedback.

Automating Security Group Updates with AWS Lambda

Post Syndicated from Ian Scofield original https://aws.amazon.com/blogs/compute/automating-security-group-updates-with-aws-lambda/

Customers often use public endpoints to perform cross-region replication or other application layer communication to remote regions. But a common problem is how do you protect these endpoints? It can be tempting to open up the security groups to the world due to the complexity of keeping security groups in sync across regions with a dynamically changing infrastructure.

Consider a situation where you are running large clusters of instances in different regions that all require internode connectivity. One approach would be to use a VPN tunnel between regions to provide a secure tunnel over which to send your traffic. A good example of this is the Transit VPC Solution, which is a published AWS solution to help customers quickly get up and running. However, this adds additional cost and complexity to your solution due to the newly required additional infrastructure.

Another approach, which I’ll explore in this post, is to restrict access to the nodes by whitelisting the public IP addresses of your hosts in the opposite region. Today, I’ll outline a solution that allows for cross-region security group updates, can handle remote region failures, and supports external actions such as manually terminating instances or adding instances to an existing Auto Scaling group.

Solution overview

The overview of this solution is diagrammed below. Although this post covers limiting access to your instances, you should still implement encryption to protect your data in transit.

If your entire infrastructure is running in a single region, you can reference a security group as the source, allowing your IP addresses to change without any updates required. However, if you’re going across the public internet between regions to perform things like application-level traffic or cross-region replication, this is no longer an option. Security groups are regional. When you go across regions it can be tempting to drop security to enable this communication.

Although using an Elastic IP address can provide you with a static IP address that you can define as a source for your security groups, this may not always be feasible, especially when automatic scaling is desired.

In this example scenario, you have a distributed database that requires full internode communication for replication. If you place a cluster in us-east-1 and us-west-2, you must provide a secure method of communication between the two. Because the database uses cloud best practices, you can add or remove nodes as the load varies.

To start the process of updating your security groups, you must know when an instance has come online to trigger your workflow. Auto Scaling groups have the concept of lifecycle hooks that enable you to perform custom actions as the group launches or terminates instances.

When Auto Scaling begins to launch or terminate an instance, it puts the instance into a wait state (Pending:Wait or Terminating:Wait). The instance remains in this state while you perform your various actions until either you tell Auto Scaling to Continue, Abandon, or the timeout period ends. A lifecycle hook can trigger a CloudWatch event, publish to an Amazon SNS topic, or send to an Amazon SQS queue. For this example, you use CloudWatch Events to trigger an AWS Lambda function that updates an Amazon DynamoDB table.

Component breakdown

Here’s a quick breakdown of the components involved in this solution:

• Lambda function
• CloudWatch event
• DynamoDB table

Lambda function

The Lambda function automatically updates your security groups, in the following way:

1. Determines whether a change was triggered by your Auto Scaling group lifecycle hook or manually invoked for a “true up” functionality, which I discuss later in this post.
2. Describes the instances in the Auto Scaling group and obtain public IP addresses for each instance.
3. Updates both local and remote DynamoDB tables.
4. Compares the list of public IP addresses for both local and remote clusters with what’s already in the local region security group. Update the security group.
5. Compares the list of public IP addresses for both local and remote clusters with what’s already in the remote region security group. Update the security group
6. Signals CONTINUE back to the lifecycle hook.

CloudWatch event

The CloudWatch event triggers when an instance passes through either the launching or terminating states. When the Lambda function gets invoked, it receives an event that looks like the following:

{
	"account": "123456789012",
	"region": "us-east-1",
	"detail": {
		"LifecycleHookName": "hook-launching",
		"AutoScalingGroupName": "",
		"LifecycleActionToken": "33965228-086a-4aeb-8c26-f82ed3bef495",
		"LifecycleTransition": "autoscaling:EC2_INSTANCE_LAUNCHING",
		"EC2InstanceId": "i-017425ec54f22f994"
	},
	"detail-type": "EC2 Instance-launch Lifecycle Action",
	"source": "aws.autoscaling",
	"version": "0",
	"time": "2017-05-03T02:20:59Z",
	"id": "cb930cf8-ce8b-4b6c-8011-af17966eb7e2",
	"resources": [
		"arn:aws:autoscaling:us-east-1:123456789012:autoScalingGroup:d3fe9d96-34d0-4c62-b9bb-293a41ba3765:autoScalingGroupName/"
	]
}

DynamoDB table

You use DynamoDB to store lists of remote IP addresses in a local table that is updated by the opposite region as a failsafe source of truth. Although you can describe your Auto Scaling group for the local region, you must maintain a list of IP addresses for the remote region.

To minimize the number of describe calls and prevent an issue in the remote region from blocking your local scaling actions, we keep a list of the remote IP addresses in a local DynamoDB table. Each Lambda function in each region is responsible for updating the public IP addresses of its Auto Scaling group for both the local and remote tables.

As with all the infrastructure in this solution, there is a DynamoDB table in both regions that mirror each other. For example, the following screenshot shows a sample DynamoDB table. The Lambda function in us-east-1 would update the DynamoDB entry for us-east-1 in both tables in both regions.

By updating a DynamoDB table in both regions, it allows the local region to gracefully handle issues with the remote region, which would otherwise prevent your ability to scale locally. If the remote region becomes inaccessible, you have a copy of the latest configuration from the table that you can use to continue to sync with your security groups. When the remote region comes back online, it pushes its updated public IP addresses to the DynamoDB table. The security group is updated to reflect the current status by the remote Lambda function.

 

Walkthrough

Note: All of the following steps are performed in both regions. The Launch Stack buttons will default to the us-east-1 region.

Here’s a quick overview of the steps involved in this process:

1. An instance is launched or terminated, which triggers an Auto Scaling group lifecycle hook, triggering the Lambda function via CloudWatch Events.
2. The Lambda function retrieves the list of public IP addresses for all instances in the local region Auto Scaling group.
3. The Lambda function updates the local and remote region DynamoDB tables with the public IP addresses just received for the local Auto Scaling group.
4. The Lambda function updates the local region security group with the public IP addresses, removing and adding to ensure that it mirrors what is present for the local and remote Auto Scaling groups.
5. The Lambda function updates the remote region security group with the public IP addresses, removing and adding to ensure that it mirrors what is present for the local and remote Auto Scaling groups.

Prerequisites

To deploy this solution, you need to have Auto Scaling groups, launch configurations, and a base security group in both regions. To expedite this process, this CloudFormation template can be launched in both regions.

Step 1: Launch the AWS SAM template in the first region

To make the deployment process easy, I’ve created an AWS Serverless Application Model (AWS SAM) template, which is a new specification that makes it easier to manage and deploy serverless applications on AWS. This template creates the following resources:

• A Lambda function, to perform the various security group actions
• A DynamoDB table, to track the state of the local and remote Auto Scaling groups
• Auto Scaling group lifecycle hooks for instance launching and terminating
• A CloudWatch event, to track the EC2 Instance-Launch Lifecycle-Action and EC2 Instance-terminate Lifecycle-Action events
• A pointer from the CloudWatch event to the Lambda function, and the necessary permissions

Download the template from here or click to launch.

Upon launching the template, you’ll be presented with a list of parameters which includes the remote/local names for your Auto Scaling Groups, AWS region, Security Group IDs, DynamoDB table names, as well as where the code for the Lambda function is located. Because this is the first region you’re launching the stack in, fill out all the parameters except for the RemoteTable parameter as it hasn’t been created yet (you fill this in later).

Step 2: Test the local region

After the stack has finished launching, you can test the local region. Open the EC2 console and find the Auto Scaling group that was created when launching the prerequisite stack. Change the desired number of instances from 0 to 1.

For both regions, check your security group to verify that the public IP address of the instance created is now in the security group.

Local region:

Remote region:

Now, change the desired number of instances for your group back to 0 and verify that the rules are properly removed.

Local region:

Remote region:

Step 3: Launch in the remote region

When you deploy a Lambda function using CloudFormation, the Lambda zip file needs to reside in the same region you are launching the template. Once you choose your remote region, create an Amazon S3 bucket and upload the Lambda zip file there. Next, go to the remote region and launch the same SAM template as before, but make sure you update the CodeBucket and CodeKey parameters. Also, because this is the second launch, you now have all the values and can fill out all the parameters, specifically the RemoteTable value.

 

Step 4: Update the local region Lambda environment variable

When you originally launched the template in the local region, you didn’t have the name of the DynamoDB table for the remote region, because you hadn’t created it yet. Now that you have launched the remote template, you can perform a CloudFormation stack update on the initial SAM template. This populates the remote DynamoDB table name into the initial Lambda function’s environment variables.

In the CloudFormation console in the initial region, select the stack. Under Actions, choose Update Stack, and select the SAM template used for both regions. Under Parameters, populate the remote DynamoDB table name, as shown below. Choose Next and let the stack update complete. This updates your Lambda function and completes the setup process.

 

Step 5: Final testing

You now have everything fully configured and in place to trigger security group changes based on instances being added or removed to your Auto Scaling groups in both regions. Test this by changing the desired capacity of your group in both regions.

True up functionality
If an instance is manually added or removed from the Auto Scaling group, the lifecycle hooks don’t get triggered. To account for this, the Lambda function supports a “true up” functionality in which the function can be manually invoked. If you paste in the following JSON text for your test event, it kicks off the entire workflow. For added peace of mind, you can also have this function fire via a CloudWatch event with a CRON expression for nearly continuous checking.

{
	"detail": {
		"AutoScalingGroupName": "<your ASG name>"
	},
	"trueup":true
}

Extra credit

Now that all the resources are created in both regions, go back and break down the policy to incorporate resource-level permissions for specific security groups, Auto Scaling groups, and the DynamoDB tables.

Although this post is centered around using public IP addresses for your instances, you could instead use a VPN between regions. In this case, you would still be able to use this solution to scope down the security groups to the cluster instances. However, the code would need to be modified to support private IP addresses.

 

Conclusion

At this point, you now have a mechanism in place that captures when a new instance is added to or removed from your cluster and updates the security groups in both regions. This ensures that you are locking down your infrastructure securely by allowing access only to other cluster members.

Keep in mind that this architecture (lifecycle hooks, CloudWatch event, Lambda function, and DynamoDB table) requires that the infrastructure to be deployed in both regions, to have synchronization going both ways.

Because this Lambda function is modifying security group rules, it’s important to have an audit log of what has been modified and who is modifying them. The out-of-the-box function provides logs in CloudWatch for what IP addresses are being added and removed for which ports. As these are all API calls being made, they are logged in CloudTrail and can be traced back to the IAM role that you created for your lifecycle hooks. This can provide historical data that can be used for troubleshooting or auditing purposes.

Security is paramount at AWS. We want to ensure that customers are protecting access to their resources. This solution helps you keep your security groups in both regions automatically in sync with your Auto Scaling group resources. Let us know if you have any questions or other solutions you’ve come up with!

Coaxing 2D platforming out of Unity

Post Syndicated from Eevee original https://eev.ee/blog/2017/10/13/coaxing-2d-platforming-out-of-unity/

An anonymous donor asked a question that I can’t even begin to figure out how to answer, but they also said anything else is fine, so here’s anything else.

I’ve been avoiding writing about game physics, since I want to save it for ✨ the book I’m writing ✨, but that book will almost certainly not touch on Unity. Here, then, is a brief run through some of the brick walls I ran into while trying to convince Unity to do 2D platforming.

This is fairly high-level — there are no blocks of code or helpful diagrams. I’m just getting this out of my head because it’s interesting. If you want more gritty details, I guess you’ll have to wait for ✨ the book ✨.

The setup

I hadn’t used Unity before. I hadn’t even used a “real” physics engine before. My games so far have mostly used LÖVE, a Lua-based engine. LÖVE includes box2d bindings, but for various reasons (not all of them good), I opted to avoid them and instead write my own physics completely from scratch. (How, you ask? ✨ Book ✨!)

I was invited to work on a Unity project, Chaos Composer, that someone else had already started. It had basic movement already implemented; I taught myself Unity’s physics system by hacking on it. It’s entirely possible that none of this is actually the best way to do anything, since I was really trying to reproduce my own homegrown stuff in Unity, but it’s the best I’ve managed to come up with.

Two recurring snags were that you can’t ask Unity to do multiple physics updates in a row, and sometimes getting the information I wanted was difficult. Working with my own code spoiled me a little, since I could invoke it at any time and ask it anything I wanted; Unity, on the other hand, is someone else’s black box with a rigid interface on top.

Also, wow, Googling for a lot of this was not quite as helpful as expected. A lot of what’s out there is just the first thing that works, and often that’s pretty hacky and imposes severe limits on the game design (e.g., “this won’t work with slopes”). Basic movement and collision are the first thing you do, which seems to me like the worst time to be locking yourself out of a lot of design options. I tried very (very, very, very) hard to minimize those kinds of constraints.

Problem 1: Movement

When I showed up, movement was already working. Problem solved!

Like any good programmer, I immediately set out to un-solve it. Given a “real” physics engine like Unity prominently features, you have two options: ⓐ treat the player as a physics object, or ⓑ don’t. The existing code went with option ⓑ, like I’d done myself with LÖVE, and like I’d seen countless people advise. Using a physics sim makes for bad platforming.

But… why? I believed it, but I couldn’t concretely defend it. I had to know for myself. So I started a blank project, drew some physics boxes, and wrote a dozen-line player controller.

Ah! Immediate enlightenment.

If the player was sliding down a wall, and I tried to move them into the wall, they would simply freeze in midair until I let go of the movement key. The trouble is that the physics sim works in terms of forces — moving the player involves giving them a nudge in some direction, like a giant invisible hand pushing them around the level. Surprise! If you press a real object against a real wall with your real hand, you’ll see the same effect — friction will cancel out gravity, and the object will stay in midair..

Platformer movement, as it turns out, doesn’t make any goddamn physical sense. What is air control? What are you pushing against? Nothing, really; we just have it because it’s nice to play with, because not having it is a nightmare.

I looked to see if there were any common solutions to this, and I only really found one: make all your walls frictionless.

Game development is full of hacks like this, and I… don’t like them. I can accept that minor hacks are necessary sometimes, but this one makes an early and widespread change to a fundamental system to “fix” something that was wrong in the first place. It also imposes an “invisible” requirement, something I try to avoid at all costs — if you forget to make a particular wall frictionless, you’ll never know unless you happen to try sliding down it.

And so, I swiftly returned to the existing code. It wasn’t too different from what I’d come up with for LÖVE: it applied gravity by hand, tracked the player’s velocity, computed the intended movement each frame, and moved by that amount. The interesting thing was that it used MovePosition, which schedules a movement for the next physics update and stops the movement if the player hits something solid.

It’s kind of a nice hybrid approach, actually; all the “physics” for conscious actors is done by hand, but the physics engine is still used for collision detection. It’s also used for collision rejection — if the player manages to wedge themselves several pixels into a solid object, for example, the physics engine will try to gently nudge them back out of it with no extra effort required on my part. I still haven’t figured out how to get that to work with my homegrown stuff, which is built to prevent overlap rather than to jiggle things out of it.

But wait, what about…

Our player is a dynamic body with rotation lock and no gravity. Why not just use a kinematic body?

I must be missing something, because I do not understand the point of kinematic bodies. I ran into this with Godot, too, which documented them the same way: as intended for use as players and other manually-moved objects. But by default, they don’t even collide with other kinematic bodies or static geometry. What? There’s a checkbox to turn this on, which I enabled, but then I found out that MovePosition doesn’t stop kinematic bodies when they hit something, so I would’ve had to cast along the intended path of movement to figure out when to stop, thus duplicating the same work the physics engine was about to do.

But that’s impossible anyway! Static geometry generally wants to be made of edge colliders, right? They don’t care about concave/convex. Imagine the player is standing on the ground near a wall and tries to move towards the wall. Both the ground and the wall are different edges from the same edge collider.

If you try to cast the player’s hitbox horizontally, parallel to the ground, you’ll only get one collision: the existing collision with the ground. Casting doesn’t distinguish between touching and hitting. And because Unity only reports one collision per collider, and because the ground will always show up first, you will never find out about the impending wall collision.

So you’re forced to either use raycasts for collision detection or decomposed polygons for world geometry, both of which are slightly worse tools for no real gain.

I ended up sticking with a dynamic body.


Oh, one other thing that doesn’t really fit anywhere else: keep track of units! If you’re adding something called “velocity” directly to something called “position”, something has gone very wrong. Acceleration is distance per time squared; velocity is distance per time; position is distance. You must multiply or divide by time to convert between them.

I never even, say, add a constant directly to position every frame; I always phrase it as velocity and multiply by Δt. It keeps the units consistent: time is always in seconds, not in tics.

Problem 2: Slopes

Ah, now we start to get off in the weeds.

A sort of pre-problem here was detecting whether we’re on a slope, which means detecting the ground. The codebase originally used a manual physics query of the area around the player’s feet to check for the ground, which seems to be somewhat common, but that can’t tell me the angle of the detected ground. (It’s also kind of error-prone, since “around the player’s feet” has to be specified by hand and may not stay correct through animations or changes in the hitbox.)

I replaced that with what I’d eventually settled on in LÖVE: detect the ground by detecting collisions, and looking at the normal of the collision. A normal is a vector that points straight out from a surface, so if you’re standing on the ground, the normal points straight up; if you’re on a 10° incline, the normal points 10° away from straight up.

Not all collisions are with the ground, of course, so I assumed something is ground if the normal pointed away from gravity. (I like this definition more than “points upwards”, because it avoids assuming anything about the direction of gravity, which leaves some interesting doors open for later on.) That’s easily detected by taking the dot product — if it’s negative, the collision was with the ground, and I now have the normal of the ground.

Actually doing this in practice was slightly tricky. With my LÖVE engine, I could cram this right into the middle of collision resolution. With Unity, not quite so much. I went through a couple iterations before I really grasped Unity’s execution order, which I guess I will have to briefly recap for this to make sense.

Unity essentially has two update cycles. It performs physics updates at fixed intervals for consistency, and updates everything else just before rendering. Within a single frame, Unity does as many fixed physics updates as it has spare time for (which might be zero, one, or more), then does a regular update, then renders. User code can implement either or both of Update, which runs during a regular update, and FixedUpdate, which runs just before Unity does a physics pass.

So my solution was:

  • At the very end of FixedUpdate, clear the actor’s “on ground” flag and ground normal.

  • During OnCollisionEnter2D and OnCollisionStay2D (which are called from within a physics pass), if there’s a collision that looks like it’s with the ground, set the “on ground” flag and ground normal. (If there are multiple ground collisions, well, good luck figuring out the best way to resolve that! At the moment I’m just taking the first and hoping for the best.)

That means there’s a brief window between the end of FixedUpdate and Unity’s physics pass during which a grounded actor might mistakenly believe it’s not on the ground, which is a bit of a shame, but there are very few good reasons for anything to be happening in that window.

Okay! Now we can do slopes.

Just kidding! First we have to do sliding.

When I first looked at this code, it didn’t apply gravity while the player was on the ground. I think I may have had some problems with detecting the ground as result, since the player was no longer pushing down against it? Either way, it seemed like a silly special case, so I made gravity always apply.

Lo! I was a fool. The player could no longer move.

Why? Because MovePosition does exactly what it promises. If the player collides with something, they’ll stop moving. Applying gravity means that the player is trying to move diagonally downwards into the ground, and so MovePosition stops them immediately.

Hence, sliding. I don’t want the player to actually try to move into the ground. I want them to move the unblocked part of that movement. For flat ground, that means the horizontal part, which is pretty much the same as discarding gravity. For sloped ground, it’s a bit more complicated!

Okay but actually it’s less complicated than you’d think. It can be done with some cross products fairly easily, but Unity makes it even easier with a couple casts. There’s a Vector3.ProjectOnPlane function that projects an arbitrary vector on a plane given by its normal — exactly the thing I want! So I apply that to the attempted movement before passing it along to MovePosition. I do the same thing with the current velocity, to prevent the player from accelerating infinitely downwards while standing on flat ground.

One other thing: I don’t actually use the detected ground normal for this. The player might be touching two ground surfaces at the same time, and I’d want to project on both of them. Instead, I use the player body’s GetContacts method, which returns contact points (and normals!) for everything the player is currently touching. I believe those contact points are tracked by the physics engine anyway, so asking for them doesn’t require any actual physics work.

(Looking at the code I have, I notice that I still only perform the slide for surfaces facing upwards — but I’d want to slide against sloped ceilings, too. Why did I do this? Maybe I should remove that.)

(Also, I’m pretty sure projecting a vector on a plane is non-commutative, which raises the question of which order the projections should happen in and what difference it makes. I don’t have a good answer.)

(I note that my LÖVE setup does something slightly different: it just tries whatever the movement ought to be, and if there’s a collision, then it projects — and tries again with the remaining movement. But I can’t ask Unity to do multiple moves in one physics update, alas.)

Okay! Now, slopes. But actually, with the above work done, slopes are most of the way there already.

One obvious problem is that the player tries to move horizontally even when on a slope, and the easy fix is to change their movement from speed * Vector2.right to speed * new Vector2(ground.y, -ground.x) while on the ground. That’s the ground normal rotated a quarter-turn clockwise, so for flat ground it still points to the right, and in general it points rightwards along the ground. (Note that it assumes the ground normal is a unit vector, but as far as I’m aware, that’s true for all the normals Unity gives you.)

Another issue is that if the player stands motionless on a slope, gravity will cause them to slowly slide down it — because the movement from gravity will be projected onto the slope, and unlike flat ground, the result is no longer zero. For conscious actors only, I counter this by adding the opposite factor to the player’s velocity as part of adding in their walking speed. This matches how the real world works, to some extent: when you’re standing on a hill, you’re exerting some small amount of effort just to stay in place.

(Note that slope resistance is not the same as friction. Okay, yes, in the real world, virtually all resistance to movement happens as a result of friction, but bracing yourself against the ground isn’t the same as being passively resisted.)

From here there are a lot of things you can do, depending on how you think slopes should be handled. You could make the player unable to walk up slopes that are too steep. You could make walking down a slope faster than walking up it. You could make jumping go along the ground normal, rather than straight up. You could raise the player’s max allowed speed while running downhill. Whatever you want, really. Armed with a normal and awareness of dot products, you can do whatever you want.

But first you might want to fix a few aggravating side effects.

Problem 3: Ground adherence

I don’t know if there’s a better name for this. I rarely even see anyone talk about it, which surprises me; it seems like it should be a very common problem.

The problem is: if the player runs up a slope which then abruptly changes to flat ground, their momentum will carry them into the air. For very fast players going off the top of very steep slopes, this makes sense, but it becomes visible even for relatively gentle slopes. It was a mild nightmare in the original release of our game Lunar Depot 38, which has very “rough” ground made up of lots of shallow slopes — so the player is very frequently slightly off the ground, which meant they couldn’t jump, for seemingly no reason. (I even had code to fix this, but I disabled it because of a silly visual side effect that I never got around to fixing.)

Anyway! The reason this is a problem is that game protagonists are generally not boxes sliding around — they have legs. We don’t go flying off the top of real-world hilltops because we put our foot down until it touches the ground.

Simulating this footfall is surprisingly fiddly to get right, especially with someone else’s physics engine. It’s made somewhat easier by Cast, which casts the entire hitbox — no matter what shape it is — in a particular direction, as if it had moved, and tells you all the hypothetical collisions in order.

So I cast the player in the direction of gravity by some distance. If the cast hits something solid with a ground-like collision normal, then the player must be close to the ground, and I move them down to touch it (and set that ground as the new ground normal).

There are some wrinkles.

Wrinkle 1: I only want to do this if the player is off the ground now, but was on the ground last frame, and is not deliberately moving upwards. That latter condition means I want to skip this logic if the player jumps, for example, but also if the player is thrust upwards by a spring or abducted by a UFO or whatever. As long as external code goes through some interface and doesn’t mess with the player’s velocity directly, that shouldn’t be too hard to track.

Wrinkle 2: When does this logic run? It needs to happen after the player moves, which means after a Unity physics pass… but there’s no callback for that point in time. I ended up running it at the beginning of FixedUpdate and the beginning of Update — since I definitely want to do it before rendering happens! That means it’ll sometimes happen twice between physics updates. (I could carefully juggle a flag to skip the second run, but I… didn’t do that. Yet?)

Wrinkle 3: I can’t move the player with MovePosition! Remember, MovePosition schedules a movement, it doesn’t actually perform one; that means if it’s called twice before the physics pass, the first call is effectively ignored. I can’t easily combine the drop with the player’s regular movement, for various fiddly reasons. I ended up doing it “by hand” using transform.Translate, which I think was the “old way” to do manual movement before MovePosition existed. I’m not totally sure if it activates triggers? For that matter, I’m not sure it even notices collisions — but since I did a full-body Cast, there shouldn’t be any anyway.

Wrinkle 4: What, exactly, is “some distance”? I’ve yet to find a satisfying answer for this. It seems like it ought to be based on the player’s current speed and the slope of the ground they’re moving along, but every time I’ve done that math, I’ve gotten totally ludicrous answers that sometimes exceed the size of a tile. But maybe that’s not wrong? Play around, I guess, and think about when the effect should “break” and the player should go flying off the top of a hill.

Wrinkle 5: It’s possible that the player will launch off a slope, hit something, and then be adhered to the ground where they wouldn’t have hit it. I don’t much like this edge case, but I don’t see a way around it either.

This problem is surprisingly awkward for how simple it sounds, and the solution isn’t entirely satisfying. Oh, well; the results are much nicer than the solution. As an added bonus, this also fixes occasional problems with running down a hill and becoming detached from the ground due to precision issues or whathaveyou.

Problem 4: One-way platforms

Ah, what a nightmare.

It took me ages just to figure out how to define one-way platforms. Only block when the player is moving downwards? Nope. Only block when the player is above the platform? Nuh-uh.

Well, okay, yes, those approaches might work for convex players and flat platforms. But what about… sloped, one-way platforms? There’s no reason you shouldn’t be able to have those. If Super Mario World can do it, surely Unity can do it almost 30 years later.

The trick is, again, to look at the collision normal. If it faces away from gravity, the player is hitting a ground-like surface, so the platform should block them. Otherwise (or if the player overlaps the platform), it shouldn’t.

Here’s the catch: Unity doesn’t have conditional collision. I can’t decide, on the fly, whether a collision should block or not. In fact, I think that by the time I get a callback like OnCollisionEnter2D, the physics pass is already over.

I could go the other way and use triggers (which are non-blocking), but then I have the opposite problem: I can’t stop the player on the fly. I could move them back to where they hit the trigger, but I envision all kinds of problems as a result. What if they were moving fast enough to activate something on the other side of the platform? What if something else moved to where I’m trying to shove them back to in the meantime? How does this interact with ground detection and listing contacts, which would rightly ignore a trigger as non-blocking?

I beat my head against this for a while, but the inability to respond to collision conditionally was a huge roadblock. It’s all the more infuriating a problem, because Unity ships with a one-way platform modifier thing. Unfortunately, it seems to have been implemented by someone who has never played a platformer. It’s literally one-way — the player is only allowed to move straight upwards through it, not in from the sides. It also tries to block the player if they’re moving downwards while inside the platform, which invokes clumsy rejection behavior. And this all seems to be built into the physics engine itself somehow, so I can’t simply copy whatever they did.

Eventually, I settled on the following. After calculating attempted movement (including sliding), just at the end of FixedUpdate, I do a Cast along the movement vector. I’m not thrilled about having to duplicate the physics engine’s own work, but I do filter to only things on a “one-way platform” physics layer, which should at least help. For each object the cast hits, I use Physics2D.IgnoreCollision to either ignore or un-ignore the collision between the player and the platform, depending on whether the collision was ground-like or not.

(A lot of people suggested turning off collision between layers, but that can’t possibly work — the player might be standing on one platform while inside another, and anyway, this should work for all actors!)

Again, wrinkles! But fewer this time. Actually, maybe just one: handling the case where the player already overlaps the platform. I can’t just check for that with e.g. OverlapCollider, because that doesn’t distinguish between overlapping and merely touching.

I came up with a fairly simple fix: if I was going to un-ignore the collision (i.e. make the platform block), and the cast distance is reported as zero (either already touching or overlapping), I simply do nothing instead. If I’m standing on the platform, I must have already set it blocking when I was approaching it from the top anyway; if I’m overlapping it, I must have already set it non-blocking to get here in the first place.

I can imagine a few cases where this might go wrong. Moving platforms, especially, are going to cause some interesting issues. But this is the best I can do with what I know, and it seems to work well enough so far.

Oh, and our player can deliberately drop down through platforms, which was easy enough to implement; I just decide the platform is always passable while some button is held down.

Problem 5: Pushers and carriers

I haven’t gotten to this yet! Oh boy, can’t wait. I implemented it in LÖVE, but my way was hilariously invasive; I’m hoping that having a physics engine that supports a handwaved “this pushes that” will help. Of course, you also have to worry about sticking to platforms, for which the recommended solution is apparently to parent the cargo to the platform, which sounds goofy to me? I guess I’ll find out when I throw myself at it later.

Overall result

I ended up with a fairly pleasant-feeling system that supports slopes and one-way platforms and whatnot, with all the same pieces as I came up with for LÖVE. The code somehow ended up as less of a mess, too, but it probably helps that I’ve been down this rabbit hole once before and kinda knew what I was aiming for this time.

Animation of a character running smoothly along the top of an irregular dinosaur skeleton

Sorry that I don’t have a big block of code for you to copy-paste into your project. I don’t think there are nearly enough narrative discussions of these fundamentals, though, so hopefully this is useful to someone. If not, well, look forward to ✨ my book, that I am writing ✨!

Low-tech Raspberry Pi robot

Post Syndicated from Rachel Churcher original https://www.raspberrypi.org/blog/low-tech-raspberry-pi-robot/

Robot-builder extraordinaire Clément Didier is ushering in the era of our cybernetic overlords. Future generations will remember him as the creator of robots constructed from cardboard and conductive paint which are so easy to replicate that a robot could do it. Welcome to the singularity.

Bare Conductive on Twitter

This cool robot was made with the #PiCap, conductive paint and @Raspberry_Pi by @clementdidier. Full tutorial: https://t.co/AcQVTS4vr2 https://t.co/D04U5UGR0P

Simple interface

To assemble the robot, Clément made use of a Pi Cap board, a motor driver, and most importantly, a tube of Bare Conductive Electric Paint. He painted the control interface onto the cardboard surface of the robot, allowing a human, replicant, or superior robot to direct its movements simply by touching the paint.

Clever design

The Raspberry Pi 3, the motor control board, and the painted input buttons interface via the GPIO breakout pins on the Pi Cap. Crocodile clips connect the Pi Cap to the cardboard-and-paint control surface, while jumper wires connect it to the motor control board.

Raspberry Pi and bare conductive Pi Cap

Sing with me: ‘The Raspberry Pi’s connected to the Pi Cap, and the Pi Cap’s connected to the inputs, and…’

Two battery packs provide power to the Raspberry Pi, and to the four independently driven motors. Software, written in Python, allows the robot to respond to inputs from the conductive paint. The motors drive wheels attached to a plastic chassis, moving and turning the robot at the touch of a square of black paint.

Artistic circuit

Clément used masking tape and a paintbrush to create the control buttons. For a human, this is obviously a fiddly process which relies on the blocking properties of the masking tape and a steady hand. For a robot, however, the process would be a simple, freehand one, resulting in neatly painted circuits on every single one of countless robotic minions. Cybernetic domination is at (metallic) hand.

The control surface of the robot, painted with bare conductive paint

One fiddly job for a human, one easy task for robotkind

The instructions and code for Clément’s build can be found here.

Low-tech solutions

Here at Pi Towers, we love seeing the high-tech Raspberry Pi integrated so successfully with low-tech components. In addition to conductive paint, we’ve seen cardboard laptops, toilet roll robots, fruit drum kits, chocolate box robots, and hamster-wheel-triggered cameras. Have you integrated low-tech elements into your projects (and potentially accelerated the robot apocalypse in the process)? Tell us about it in the comments!

 

The post Low-tech Raspberry Pi robot appeared first on Raspberry Pi.

Browser hacking for 280 character tweets

Post Syndicated from Robert Graham original http://blog.erratasec.com/2017/09/browser-hacking-for-280-character-tweets.html

Twitter has raised the limit to 280 characters for a select number of people. However, they left open a hole, allowing anybody to make large tweets with a little bit of hacking. The hacking skills needed are basic hacking skills, which I thought I’d write up in a blog post.


Specifically, the skills you will exercise are:

  • basic command-line shell
  • basic HTTP requests
  • basic browser DOM editing

The short instructions

The basic instructions were found in tweets like the following:
These instructions are clear to the average hacker, but of course, a bit difficult for those learning hacking, hence this post.

The command-line

The basics of most hacking start with knowledge of the command-line. This is the “Terminal” app under macOS or cmd.exe under Windows. Almost always when you see hacking dramatized in the movies, they are using the command-line.
In the beginning, the command-line is all computers had. To do anything on a computer, you had to type a “command” telling it what to do. What we see as the modern graphical screen is a layer on top of the command-line, one that translates clicks of the mouse into the raw commands.
On most systems, the command-line is known as “bash”. This is what you’ll find on Linux and macOS. Windows historically has had a different command-line that uses slightly different syntax, though in the last couple years, they’ve also supported “bash”. You’ll have to install it first, such as by following these instructions.
You’ll see me use command that may not be yet installed on your “bash” command-line, like nc and curl. You’ll need to run a command to install them, such as:
sudo apt-get install nc curl
The thing to remember about the command-line is that the mouse doesn’t work. You can’t click to move the cursor as you normally do in applications. That’s because the command-line predates the mouse by decades. Instead, you have to use arrow keys.
I’m not going to spend much effort discussing the command-line, as a complete explanation is beyond the scope of this document. Instead, I’m assuming the reader either already knows it, or will learn-from-example as we go along.

Web requests

The basics of how the web works are really simple. A request to a web server is just a small packet of text, such as the following, which does a search on Google for the search-term “penguin” (presumably, you are interested in knowing more about penguins):
GET /search?q=penguin HTTP/1.0
Host: www.google.com
User-Agent: human
The command we are sending to the server is GET, meaning get a page. We are accessing the URL /search, which on Google’s website, is how you do a search. We are then sending the parameter q with the value penguin. We also declare that we are using version 1.0 of the HTTP (hyper-text transfer protocol).
Following the first line there are a number of additional headers. In one header, we declare the Host name that we are accessing. Web servers can contain many different websites, with different names, so this header is usually imporant.
We also add the User-Agent header. The “user-agent” means the “browser” that you use, like Edge, Chrome, Firefox, or Safari. It allows servers to send content optimized for different browsers. Since we are sending web requests without a browser here, we are joking around saying human.
Here’s what happens when we use the nc program to send this to a google web server:
The first part is us typing, until we hit the [enter] key to create a blank line. After that point is the response from the Google server. We get back a result code (OK), followed by more headers from the server, and finally the contents of the webpage, which goes on from many screens. (We’ll talk about what web pages look like below).
Note that a lot of HTTP headers are optional and really have little influence on what’s going on. They are just junk added to web requests. For example, we see Google report a P3P header is some relic of 2002 that nobody uses anymore, as far as I can tell. Indeed, if you follow the URL in the P3P header, Google pretty much says exactly that.
I point this out because the request I show above is a simplified one. In practice, most requests contain a lot more headers, especially Cookie headers. We’ll see that later when making requests.

Using cURL instead

Sending the raw HTTP request to the server, and getting raw HTTP/HTML back, is annoying. The better way of doing this is with the tool known as cURL, or plainly, just curl. You may be familiar with the older command-line tools wget. cURL is similar, but more flexible.
To use curl for the experiment above, we’d do something like the following. We are saving the web page to “penguin.html” instead of just spewing it on the screen.
Underneath, cURL builds an HTTP header just like the one we showed above, and sends it to the server, getting the response back.

Web-pages

Now let’s talk about web pages. When you look at the web page we got back from Google while searching for “penguin”, you’ll see that it’s intimidatingly complex. I mean, it intimidates me. But it all starts from some basic principles, so we’ll look at some simpler examples.
The following is text of a simple web page:
<html>
<body>
<h1>Test</h1>
<p>This is a simple web page</p>
</body>
</html>
This is HTML, “hyper-text markup language”. As it’s name implies, we “markup” text, such as declaring the first text as a level-1 header (H1), and the following text as a paragraph (P).
In a web browser, this gets rendered as something that looks like the following. Notice how a header is formatted differently from a paragraph. Also notice that web browsers can use local files as well as make remote requests to web servers:
You can right-mouse click on the page and do a “View Source”. This will show the raw source behind the web page:
Web pages don’t just contain marked-up text. They contain two other important features, style information that dictates how things appear, and script that does all the live things that web pages do, from which we build web apps.
So let’s add a little bit of style and scripting to our web page. First, let’s view the source we’ll be adding:
In our header (H1) field, we’ve added the attribute to the markup giving this an id of mytitle. In the style section above, we give that element a color of blue, and tell it to align to the center.
Then, in our script section, we’ve told it that when somebody clicks on the element “mytitle”, it should send an “alert” message of “hello”.
This is what our web page now looks like, with the center blue title:
When we click on the title, we get a popup alert:
Thus, we see an example of the three components of a webpage: markup, style, and scripting.

Chrome developer tools

Now we go off the deep end. Right-mouse click on “Test” (not normal click, but right-button click, to pull up a menu). Select “Inspect”.
You should now get a window that looks something like the following. Chrome splits the screen in half, showing the web page on the left, and it’s debug tools on the right.
This looks similar to what “View Source” shows, but it isn’t. Instead, it’s showing how Chrome interpreted the source HTML. For example, our style/script tags should’ve been marked up with a head (header) tag. We forgot it, but Chrome adds it in anyway.
What Google is showing us is called the DOM, or document object model. It shows us all the objects that make up a web page, and how they fit together.
For example, it shows us how the style information for #mytitle is created. It first starts with the default style information for an h1 tag, and then how we’ve changed it with our style specifications.
We can edit the DOM manually. Just double click on things you want to change. For example, in this screen shot, I’ve changed the style spec from blue to red, and I’ve changed the header and paragraph test. The original file on disk hasn’t changed, but I’ve changed the DOM in memory.
This is a classic hacking technique. If you don’t like things like paywalls, for example, just right-click on the element blocking your view of the text, “Inspect” it, then delete it. (This works for some paywalls).
This edits the markup and style info, but changing the scripting stuff is a bit more complicated. To do that, click on the [Console] tab. This is the scripting console, and allows you to run code directly as part of the webpage. We are going to run code that resets what happens when we click on the title. In this case, we are simply going to change the message to “goodbye”.
Now when we click on the title, we indeed get the message:
Again, a common way to get around paywalls is to run some code like that that change which functions will be called.

Putting it all together

Now let’s put this all together in order to hack Twitter to allow us (the non-chosen) to tweet 280 characters. Review Dildog’s instructions above.
The first step is to get to Chrome Developer Tools. Dildog suggests F12. I suggest right-clicking on the Tweet button (or Reply button, as I use in my example) and doing “Inspect”, as I describe above.
You’ll now see your screen split in half, with the DOM toward the right, similar to how I describe above. However, Twitter’s app is really complex. Well, not really complex, it’s all basic stuff when you come right down to it. It’s just so much stuff — it’s a large web app with lots of parts. So we have to dive in without understanding everything that’s going on.
The Tweet/Reply button we are inspecting is going to look like this in the DOM:
The Tweet/Reply button is currently greyed out because it has the “disabled” attribute. You need to double click on it and remove that attribute. Also, in the class attribute, there is also a “disabled” part. Double-click, then click on that and removed just that disabled as well, without impacting the stuff around it. This should change the button from disabled to enabled. It won’t be greyed out, and it’ll respond when you click on it.
Now click on it. You’ll get an error message, as shown below:
What we’ve done here is bypass what’s known as client-side validation. The script in the web page prevented sending Tweets longer than 140 characters. Our editing of the DOM changed that, allowing us to send a bad request to the server. Bypassing client-side validation this way is the source of a lot of hacking.
But Twitter still does server-side validation as well. They know any client-side validation can be bypassed, and are in on the joke. They tell us hackers “You’ll have to be more clever”. So let’s be more clever.
In order to make longer 280 characters tweets work for select customers, they had to change something on the server-side. The thing they added was adding a “weighted_character_count=true” to the HTTP request. We just need to repeat the request we generated above, adding this parameter.
In theory, we can do this by fiddling with the scripting. The way Dildog describes does it a different way. He copies the request out of the browser, edits it, then send it via the command-line using curl.
We’ve used the [Elements] and [Console] tabs in Chrome’s DevTools. Now we are going to use the [Network] tab. This lists all the requests the web page has made to the server. The twitter app is constantly making requests to refresh the content of the web page. The request we made trying to do a long tweet is called “create”, and is red, because it failed.
Google Chrome gives us a number of ways to duplicate the request. The most useful is that it copies it as a full cURL command we can just paste onto the command-line. We don’t even need to know cURL, it takes care of everything for us. On Windows, since you have two command-lines, it gives you a choice to use the older Windows cmd.exe, or the newer bash.exe. I use the bash version, since I don’t know where to get the Windows command-line version of cURL.exe.
There’s a lot of going on here. The first thing to notice is the long xxxxxx strings. That’s actually not in the original screenshot. I edited the picture. That’s because these are session-cookies. If inserted them into your browser, you’d hijack my Twitter session, and be able to tweet as me (such as making Carlos Danger style tweets). Therefore, I have to remove them from the example.
At the top of the screen is the URL that we are accessing, which is https://twitter.com/i/tweet/create. Much of the rest of the screen uses the cURL -H option to add a header. These are all the HTTP headers that I describe above. Finally, at the bottom, is the –data section, which contains the data bits related to the tweet, especially the tweet itself.
We need to edit either the URL above to read https://twitter.com/i/tweet/create?weighted_character_count=true, or we need to add &weighted_character_count=true to the –data section at the bottom (either works). Remember: mouse doesn’t work on command-line, so you have to use the cursor-keys to navigate backwards in the line. Also, since the line is larger than the screen, it’s on several visual lines, even though it’s all a single line as far as the command-line is concerned.
Now just hit [return] on your keyboard, and the tweet will be sent to the server, which at the moment, works. Presto!
Twitter will either enable or disable the feature for everyone in a few weeks, at which point, this post won’t work. But the reason I’m writing this is to demonstrate the basic hacking skills. We manipulate the web pages we receive from servers, and we manipulate what’s sent back from our browser back to the server.

Easier: hack the scripting

Instead of messing with the DOM and editing the HTTP request, the better solution would be to change the scripting that does both DOM client-side validation and HTTP request generation. The only reason Dildog above didn’t do that is that it’s a lot more work trying to find where all this happens.
Others have, though. @Zemnmez did just that, though his technique works for the alternate TweetDeck client (https://tweetdeck.twitter.com) instead of the default client. Go copy his code from here, then paste it into the DevTools scripting [Console]. It’ll go in an replace some scripting functions, such like my simpler example above.
The console is showing a stream of error messages, because TweetDeck has bugs, ignore those.
Now you can effortlessly do long tweets as normal, without all the messing around I’ve spent so much text in this blog post describing.
Now, as I’ve mentioned this before, you are only editing what’s going on in the current web page. If you refresh this page, or close it, everything will be lost. You’ll have to re-open the DevTools scripting console and repaste the code. The easier way of doing this is to use the [Sources] tab instead of [Console] and use the “Snippets” feature to save this bit of code in your browser, to make it easier next time.
The even easier way is to use Chrome extensions like TamperMonkey and GreaseMonkey that’ll take care of this for you. They’ll save the script, and automatically run it when they see you open the TweetDeck webpage again.
An even easier way is to use one of the several Chrome extensions written in the past day specifically designed to bypass the 140 character limit. Since the purpose of this blog post is to show you how to tamper with your browser yourself, rather than help you with Twitter, I won’t list them.

Conclusion

Tampering with the web-page the server gives you, and the data you send back, is a basic hacker skill. In truth, there is a lot to this. You have to get comfortable with the command-line, using tools like cURL. You have to learn how HTTP requests work. You have to understand how web pages are built from markup, style, and scripting. You have to be comfortable using Chrome’s DevTools for messing around with web page elements, network requests, scripting console, and scripting sources.
So it’s rather a lot, actually.
My hope with this page is to show you a practical application of all this, without getting too bogged down in fully explaining how every bit works.