Fork me on GitHub
#ldnclj
<
2015-10-13
>
maleghast06:10:33

Anyone prepared to share what they are doing about repeatability and configuration on AWS..? Clearly OpsWorks is already there, and that’s Chef by another name as far as I can tell, but it’s clear that Puppet and Ansible are both becoming or are already very AWS friendly… I am planning a personal development that I want to be able to orchestrate, but I am new to doing the DevOps Toolchain for myself and I am hoping someone is prepared to state a preference based on having walked this road ahead of me..?

maleghast06:10:58

Is Pallet still a thing, and should I be looking at that?

maleghast06:10:54

Or should I just embrace ECS and Docker and go that route, or combine the two..? Can you tell that there seems to me an embarrassment of riches and I have decision paralysis..? 😉

maleghast06:10:31

(Clearly I am planning to build a Clojure app, so I am asking in that context 😉 )

pupeno06:10:08

Good morning.

agile_geek06:10:04

@maleghast: been thinking about exactly that for over a year but haven’t had chance to really work out what is best. I know ppl working on Spark on AWS using Mesos but that plays into a slightly different problem space.

maleghast06:10:44

@agile_geek: Yeah, that’s a slightly different and yet at the same time remarkably similar problem domain

gjnoonan06:10:06

Morning all

agile_geek06:10:10

I get the feeling Pallet is a bit quiet…not sure it’s dead but not sure anyone is actively doing much on it atm.

maleghast06:10:45

And if I were thinking of a system / platform that needed 100s of containers or EC2s, then Mesos might just be the ticket actually, but it feels too heavy for what I need.

maleghast06:10:25

@agile_geek: Yeah, I was concerned that Pallet had gone “a bit quiet” - I know a couple of people who were all-about-Pallet, early on, but I don’t have a current temperature check from them on it.

agile_geek06:10:03

@maleghast: Mesos is less about deployment/delivery and more about orchestration and resource management. You’re right, if you containerise a lot then Mesos may come back into play. And certainly if you’re working with services you need to flex fast.

gjnoonan06:10:34

maleghast: I use terraform + packer on core os for repeat configuration, with Kubernetes for orchestration of containers trying to figure out the best way to get deis integrated with Kubernetes too

maleghast06:10:58

@gjnoonan: OK, there are two technologies in there that I have heard of but know nothing about (Terraform and Kubernetes) and two that I have next-to-no knowledge on (Deis and Packer), so I will do some Googling / reading and see if it’s going to do what I want 😉 Thanks

agile_geek06:10:06

@gjnoonan: fancy submitting a talk to #C075TNSSC about any of them? I, for one, would be interested.

gjnoonan06:10:15

@agile_geek I had though about it, or a talk on "the immutable law firm - using clojure and datomic to shape the law" (title wip) but for personal reasons I currently don't know ow if I can attend let alone talk 😔

agile_geek06:10:00

@gjnoonan: shame. Sounds like a great subject!

gjnoonan06:10:36

Maybe next year once I've got all this stuff sorted, or euroclojure

maleghast07:10:14

@gjnoonan: Yeah, I’d want to see that talk (on Vimeo after the conf, as I’m not coming back for Clojure confs from SE Asia 😞 )

maleghast07:10:50

I have to go into two meetings that are happening in parallel - FML 😞

thomas07:10:40

Morning folks

gjnoonan08:10:50

morning thomas

maleghast09:10:26

Good morning simple_smile

xlevus09:10:00

Good talks last night

xlevus09:10:53

totes going to ditch the "I don't know what this does, why doesn't this work, oh god stacktrace from hell" luminus template in favour of that framework that I can't remember the name of

jamiei11:10:20

xlevus: Was it liberator?

xlevus11:10:35

no. It was pedestal

jamiei11:10:22

That makes sense 😄

tcrayford11:10:37

@xlevus: sadly I think a lot of clojure programming revolves around "stacktrace from hell" 😞

xlevus11:10:22

I have noticed this. I think Ring makes it extra bad as everything's called in a stack of middleware

mccraigmccraig11:10:21

xlevus: you should try aleph - the deferred callback-chain stacktraces make ring's look sweet and compact

malcolmsparks13:10:06

stack traces will always be tied to threads, and become less useful in an async context. That's why I'm looking for alternatives, like @stathissideris 's new project positano.

malcolmsparks13:10:33

don't forget @davidhumphreys tip of 'named anonymous functions' 😉

malcolmsparks13:10:47

i.e.: (fn fred [x] ...)

stathissideris13:10:09

@malcolmsparks: it's ready for your use case btw, see email simple_smile

stathissideris13:10:44

needs more documentation, coming in the next few days

malcolmsparks13:10:56

@stathissideris: it's on the todo list 😉

stathissideris13:10:38

cool, looking forward to any flaws to being uncovered!

malcolmsparks13:10:31

Unfortunately my commute time has been monopolised by my attempt to solve what seemed to be an easy problem - async handling of incoming multipart/form-data streams. I'm now halfway down a rabbit hole of boundary detection over straddling buffers and stateful transducers...

malcolmsparks13:10:19

I feel like I'm resurfacing now, but useful personal reminder that my own skills at estimating task size are still woeful

stathissideris13:10:04

@malcolmsparks: I'm halfway down the rabbit hole of program analysis myself, spent the past few days reading papers of how datalog is used for that sort of task, and studying the code of clj-refactor, sleight, clojure.tools.analyze etc. Fun, but I'm itching to write some real code now...