#dynolocker - February 2019

@Erik Osterman (Cloud Posse) has joined the channel

set the channel description: https://github.com/joshmyers/dynolocker

@joshmyers has joined the channel

@mumoshu has joined the channel

@joshmyers was just talking with @mumoshu about your tool

i've been referring a bunch of people over to dynolocker

there's no good solution like this that does distributed locking.

@joshmyers hi!

@mumoshu has an interesting (common) use-case about renewing locks

@Andriy Knysh (Cloud Posse) has joined the channel

@joshmyers is based in GMT (London), so he'll be online tomorrow

@Michal Matyjek has joined the channel

@Michal Matyjek are you guys doing any distributed locking?

@dustinvb has joined the channel

I don't remember

I was also sharing dynolocking with @dustinvb from codefresh

ah i remembered that i had created a very similar command myself 3 years ago, but in ruby
https://github.com/crowdworks/joumae-ruby#cli

i definitely like a command written in golang 🙂 am eager to submit a pr to dynolocker if that makes sense

yea, let's breath some life into it

also wire it up with a codefresh pipeline to build binary releases

then we can also distribute through cloudposse/packages

so the use case is...?

see the snippet above

basically, you need to lock something which could be used across pipelines or even within the same pipeline

for example, if you're terraforming, you want to lock a project to a pull request

(e.g. like in atlantis)

also, i've had the problem we're we merge 2 PRs too close to each other

and we end up doing concurrent helm deployments of the same app

we should really be locking a helm release before performing a helm deployment

that's (2) very strong use-cases for locking

set the channel topic: https://github.com/joshmyers/dynolocker

probably it is unlikely to happen when you run the pipeline on a new commit to e.g. master

(happened to me yesterday... and we have wait: true, so second release was rolling out destroying pods from first release)

first release then was blocked by second release completing

but i do have encountered the same situation as erik’s in my terraform pipeline powered by github flow(the latest one, apply/deploy before merging for final testing)

i think for terraform it's scarier

are you using codefresh with terraform now?

not yet, but i’m seriously considering to move to codefresh for anything related to github flow

Morning 👋

Sure, PRs always welcome

it was a hacky tool whipped up due to a race condition when bootstrapping Vault nodes (a few years back now)

locksmithctl by coreOS may also be interesting

That one requires etcd, no?

Yeah it was based off etcd

That might not work well if running outside of k8s

dynolocker might not work well outside of AWS

well, def won’t 🙂

Haha well, kinda depends on what “outside” means ;-)

But yea, I get your point. For our purposes, AWS is an acceptable requirement.

(And already use it for terraform locking)

Yeah, I was going to try and do more with it and actually have it do sane things with process and signals etc but never got around to it

huh, I like this Variant 😄

@richwine has joined the channel

set the channel description: https://github.com/joshmyers/dynolocker
Archive: https://archive.sweetops.com/dynolocker/