Fork me on GitHub
#clojure-uk
<
2017-01-19
>
ricardo08:01:35

Thanks @agile_geek 🙂 I only visit occasionally, though, so I’ll probably be quiet around here.

ricardo08:01:55

I will be in town for Monikigras next week, in case anyone’s attending.

maleghast08:01:13

'Ello all - it's late afternoon here of course, but "morning!" anyway

maleghast09:01:47

I am relieved to hear that staying with running our own Jenkins server rather than spending 4x the cost on TravisCI or CircleCI turns out to be a great idea... Mostly...

maleghast09:01:12

It is an SPF on our stack / platform and that worries me...

agile_geek09:01:10

otfrom: are you runnig your own CI servers now? AWS and Jenkins?

glenjamin09:01:27

@maleghast do you have any conf management or automation on jenkins? At a former job we got it into a state where we could delete the master and rebuild it in a working state by reprovisioning in chef

maleghast09:01:45

@glenjamin - We were just__ talking about that here in the office. We are not great at automation - about 35% of our stack / platform is repeatable and the CI box is not one of the things that is. We are going to have a group chat tomorrow, maybe Monday on a list of things that we need to get done for long term maintainability and that's right at the top of the list.

maleghast09:01:22

I reckon we could put all the Jenkins config in a Private S3 bucket and write / modify an Ansible playbook to build a jenkins instance and apply it.

maleghast09:01:42

Ooh, I will go take a look at that now... That might be even better.

glenjamin09:01:06

the most reliable approach i’ve found is to configure jenkins via the UI, then export the job config back into ansible or equivalent as a template

maleghast09:01:20

After all the build history is not something we care about all that much unless there is a problem so if we can make the config and jobs portable then that would be great.

maleghast09:01:09

That plugin looks like a winner moving forward at least. Thanks @glenjamin

gjnoonan09:01:38

Currently using Gitlab CI as a core piece of our CI/CD pipeline.

dominicm09:01:40

@gjnoonan Are you using http://gitlab.com? Or a privately hosted version.

gjnoonan09:01:07

@dominicm privately hosted

otfrom09:01:25

agile_geek I'm just a manager nowadays. You want to ask @elise_huard @tcoupland and @acron about that

tcoupland09:01:41

we've got jenkins running as a DCOS package, so it makes use of the mesos cluster for running jobs. Backup are the crucial thing to get sorted, also save yourself a headache a pin the jenkins master to a node. We're using SlimBackup and then a custom job to upload the backups to S3. When the cluster is getting restored the backup is retrieved and then using a hero script we stuff the backup onto the new jenkins server and restore it.

maleghast09:01:12

@otfrom "I'm just a manager nowadays" - say it ain't so! 😉

maleghast09:01:09

@tcoupland - Sounds interesting, but we don't have the need of the heft that Mesos brings to the table, not by a long shot.

tcoupland10:01:32

@maleghast that sounds like a very sensible decision

maleghast10:01:25

Well, thanks... If I have to run 100s or 1000s of instances at some point in the future I may well get my Mesos cherry popped, but for right now we can achieve a lot without getting into that. One of my guys is keen to explore K8s instead of ECS and I am broadly in agreement, but building a K8s cluster inside an existing VPC is "complicated" and getting separate VPCs to talk to one another (this is all on AWS) is not as trivial as it might or could be either, so I am hoping that we are not going to switch until we are ready to take down the legacy VPC and entirely provision a new one with a K8s cluster inside it, along with the other bits and pieces...

gjnoonan10:01:12

I run a couple of k8s clusters, it's great. Also some "legacy" ones manually put together with Consul and some Magic lol

agile_geek10:01:38

@otfrom I've been a manager since 97!

mccraigmccraig13:01:37

hey south-easterly people - i'm looking to hire a clj+cljs dev - if you know of anyone who might be interested please send them in my direction or here https://www.yapster.info/join-us

glenjamin13:01:47

@maleghast A friend of mine works for http://convox.com/ which may be of interest

glenjamin13:01:03

it’s a heroku-style thing made by some early heroku employees that runs your stuff on ECS

maleghast14:01:04

@glenjamin Thanks I will take a look 🙂

yogidevbear14:01:23

Afternoon ☀️

otfrom16:01:20

maleghast I changed a regexp in some sparkling code yesterday and ran a spark job. So, not completely

maleghast17:01:07

@otfrom - Thank goodness!

otfrom18:01:15

maleghast no code today. I might be pure manager now. We'll see what happens tomorrow

maleghast18:01:44

@otfrom - At least if you go that way you know what you are missing 😉

otfrom18:01:52

will I? won't I forget all about joy and become grey?

otfrom18:01:13

😨 TOO LATE!

agile_geek20:01:08

@otfrom at least you have hair though...I'm missing out grey and going straight to white