ryanpers

NT = neurotypical, standard lingo in "aspy" forums. You're probably pretty normal. Expect eye contact and all.

On a technical note, it's not clear to me how you KNOW that chef or puppet has solved your system state problem. Now if its limited to doing super basic stuff, ok great, but then you really arent adding enough value to justify the huge cost of these systems.

I'd also like to hear what you have to say about "throw software on top of it"... for example, I want all 40 webservers to have the same build revision of our codebase. But I need positive feedback - and a gradual rollout possibly. Eventually run puppet clients really dont tell us this. Oh wait, I mean I want to rollback to the previous deployed version, because it's broken horribly.

These practical software deployment issues I find just are not addressed by the system configuration management tools like chef, puppet.

And that is why they are so passionate about it, because a million people have told them 'just use chef' and yet chef doesnt solve their problem at all, nor does it solve any fraction of their problem either.

One of the biggest problems as I noted in a different comment is that chef doesnt solve _software_ deployment. Configuration management and "best effort file pushes" is not really appropriate for running a website. Sure you could do something with chef, but waiting 20 minutes for a rollback would be pretty lame-o.

The issue here is software push and service orchestration. How do you do a rolling restart of a cluster in an automated fashion?

As for 'configuration management', I feel like puppet, chef, etc arent catch all solutions for 'i want my config to be the same on all machines' for the above mentioned 20 minute cycle.

leaky abstractions. It's important, but there is also a cargo cultism of asking questions in this certain way that is imitated in form only from how people envision google demands their engineers.

bad interview process... but they are correct in the sense that chef and puppet solve problems, but that problem is NOT software and clustered software deployment. You could hack it on top of it sure, but it'd be more straight forward to basically start from scratch. Hell bash for ssh loops would work better.

Perhaps they really were looking for someone who already thinks they way they do, and that is valid. they probably dont need you in their office for a few weeks saying "i could do this in chef if..." until they fired you for "poor fit".

Also dont forget that tech/aspy folks just are... blunt in a way that most people interpret as "insensitive". Maybe they're just being efficient? You write a blog, seem pretty eloquent, so I'm guessing you are NT, so perhaps not a good fit after all.

source code management is way beyond 'a computer utility', but since you brought it up, here are some other computer utilities:

- dd - bash - find

Most of those require quite a bit of time to master.

Now, moving past that, git isn't merely a computer utility, its a foundational element in programming. It's actually more primitive than a programming language itself! The thing that separates the high school students from the "professional" programmers is SCM, and mastery of it. Having seen hardcore dev done on 10m loc code bases in perforce, you need some perspective on how hard git is and isnt.

One of the biggest problems I find is that people attempt to understand git via the lens of SVN. Once you free yourself from the past and understand the basic fundamentals of git, and cast aside your svn-isms you will be zen.

I for one wouldn't hire a developer who isnt intimately familiar with at least one SCM system. Zip files named v1, v2, v3 anyone?

Accumulo is so similar to HBase, that its major feature sets is merely an application that can be implemented on top of HBase.

Interesting article, but its only skin deep... The larger issues are down the road, it's where the engineers and the sophomore programmers are separated.

its the futex() call that is bugged, so therefore things that are highly multi-threaded... for example mysql and java!

Many unix things are not as multithreaded as those examples.

my account got suspended for 24 hours. its very bad. no recourse is the biggest problem. I had 2 phase auth turned on, dont spam, etc, etc.

google = an algorithm stomping on a human face.. forever!

Scala is quite complex, not just the base language, but the standard libraries, the ability to comprehend code by just looking at the lines (implicits!), and the eventual full implications of features as carried out to the logical conclusion by std libs, etc.

I wrote a lengthy email describing a year's worth of experience to scala, and here is an excerpt:

If we think that one of the goals of our developers (jr, mid and even sr) is to fully understand as much of their execution environment and stack as possible, the sheer bulk of the language and its standard libraries becomes worrysome. I could probably re-write java.util.concurrent and java collections from whole cloth if necessary, but I could NOT re-write the Scala collection libraries.

I believe in team building over lunch, but free food is a terrible attractant and retention item.

People who are intrinsically motivated to do excellent work might be motivated to join due to food, but they really join for autonomy, mastery and purpose. Free food is none of these things. And ultimately you end up attracting people who are after the perks, and may not end up contributing to a passionate environment.

If you view a computer science degree, as teaching you particular sub-technologies (eg: CORBA, HTTP, etc), then sure, it is inevitable that one thinks of a CS degree as a time-limited paper slip.

This is the very definition of vocational training.

But there is so much more to computer science, and technology than vocational training. I hate to say it, but your math classes and algorithm analysis classes? They were some of the most important classes you took. Even classes like compilers are not invalid, even though we dont quite build compilers in the same way as the dragon book says we should. But some things are the same, and the nature of them does not change as fast as people think. For example, parsing has a number of approaches, but they are all inter-related, and learning about one in depth lets you understand why that approach is not ideal or what kinds of tradeoffs one is making.

Another example... as much as operating systems have evolved, many aspects of them just .. have... not ... changed. EG: IPC in modern Unix systems. Pretty much all the same as 1995 (the first year I coded sockets).

This is pretty much the general education vs vocational argument all over again.

Having used Tmobile for years, AT&T is NOT the worst carrier. Even here in SF, AT&T is better than tmobile in terms of indoor data delivery.

The notion of replacing the sun java6 with a openjdk7 is extremely laughable. If you have a high performance java server app, openjdk just doesnt cut it.

As a developer who had need to run high performance Java, I gotta say your #1 option is just not an option. OpenJDK with icedtea isn't even remotely close to a replacement.

I understand that Oracle is forcing your hand, but the lack of compassion and sympathy and the ignorant insulting "recommendations" is really off putting.

I really like this comment. The MBA-ization of corporations is not a thing to be praised. The reason why companies fail is because management is disconnected from the realities of their business and manage it from a spreadsheet point of view. See "Execution: The discipline of getting things done" for a lucid explanation of what execution is, why it matters and why companies succeed and fail. Hint: Companies who have CEOs that know little of how the company actually works are not to be emulated.

I always find the gushing articles about Amazon super annoying as an ex-amazonian.

Part of amazon's business practice is to take people, chew them up, spit them out. Replace them with new grads.

Regardless of if this is successful or not, the question really should be is this ACCEPTABLE or not.

After all, there are many successful approaches that are illegal and immoral and we eschew them and universally revile them.

The most apparent thing from this was the author and I are not in the same business at all. I work for startups, technology and all that. But without a doubt all the big stuff happens in a small part of the world. Unique and important major technology innovations. Not just more social media apps. Social media apps are the applications the Internet was built for. They are more akin to HTML websites circa 1996.

Maybe in a world where you have disposable software products. But if you are developing something more complex, you can have a substantial period where you are developing science, technology and you cant bootstrap that on ramen when the people you might need to pay for it all can get real bucks elsewhere doing the same thing.

Then again, HN is very noddy-apps-oriented.

so you take the requirements from the customers to the developers?

So in theory you are correct, but the reality on the ground is that 96gb in a box is getting to be pretty standard now.

Also the other thing to remember is that the standard java GCs dont reliably work under all workloads above 2-4GB. You can make a 8gb or 16gb heap work, but if you hit a full-compaction your JVM will go MIA for MINUTES at a time. So you can't even scale to a 16gb ram heap.

The 512GB is for the previous customers who do in fact use heaps of that size.

Always ask yourself before you send an email, what are you hoping to do with the email? Get something out of someone? Maintain or build a relationship? Make that person feel bad? Gloat?

In this case it sounds like you had no purpose at all... Just "being nice". It was probably interpreted as "you are trying to make me feel bad" by your ex-boss. Note I did not say MISINTERPRETED, because that implies it is his fault. It is not, it is up to you to control how you appear and everything surrounding that messaging.

You learned a hard lesson here, you lost a potential future ally. Shrug, learn the lesson and move on.

And finally, it ie never improper to offer ex-coworkers jobs, just be aware of: - legal obligations wrt non-compete, and IP. it might be wise to honor an non-compete even if it isnt legally enforceable. Don't like non-competes? dont sign them. - relationship status and feeling. how will the other side feel?

probably because at nearly every place I've worked at the sysadmin/ops group was the #1 obstacle to getting things done. Sure some of it was "saving you from yourself", but a lot of was pure obstinatance for no other reason than I think sysadmins just cant turn it off anymore... much as your mom says "dont frown your face might get stuck", it appears a lot of people (not just sysadmins, developers too - you know the one, in the corner who is old and surly, been there forever, and hasnt been promoted in about as long).

The deck is stacked against sysadmins - developers are the people who deliver business value to leaders, and sysadmins are the ones who cost money and say 'no'. When you are on a tight timetable, the person who says "sure" is the one who gets the high fives... assuming things work out, of course.

wtf?

Locally in Seattle, Amazon has a really bad rep. Only on places like hackernews where everyone is excited about AWS (despite its many, huge, annoying flaws, remember when ebs when down - AGAIN?) do people not realize this.

Amazon has lost a lot of key tech talent, it's been happening since about 2006 when the economy picked up a lot more.

There is no effective Seattle startup culture. I worked in the Seattle area for 7 years, and startups are not really a big thing. Going to work at microsoft is a big thing. You make more money there anyways.

A lot of this is due to the lack of a VC infrastructure I think. no sand hill road there.

when i left amazon in 2006 that was MOST DEF not true. well not the backendy parts of amazon. maybe some of the webservers were, but i somehow doubt it.

because you dont fix everything with another layer of indirection. add in some more queuing (which as we know never has a problem) just to ger around 'no threads' seems stupid.

ted you're so cool be my friend pleaseeeee

dboza is pretty awesome, beck didnt really address the underlying issue. If you have any real computation, node.js is not your solution. I figured out this in event-driven IO back when I was in school 15 years ago fffffff