<@UB89GSWCF> <github.com/cloudposse/atmos/pull/299...
# atmos-dev
e
@Andriy Knysh (Cloud Posse) github.com/cloudposse/atmos/pull/2992
👀 1
merged 1
z
Oh nice. This goes in the atmos.yaml?
We needed this with retries. Multiple pipelines trying to apply the same stack/component. Just implemented it using TF vars
e
Yes, have the same thing planned for retries
z
Sweet, yeah so if we hit a lock error hopefully we can do some kind of exponential backoff thing. We've basically consolidated all of our workflows down to a single DAG workflow now
e
Yes, they can be combined.
Add the automatic lock timeouts in terraform; then add the retries as an extra measure
👍 1
z
Sounds good. Yeah next thing I'm going to look into is all the new cache thing you guys added
Our network we have to egress out of has been exceptionally flakey for the past month lol
e
Yes, that's pretty rad. Plus it can be combined with the native github caching, storing the cache as an artifact
z
Randomly you hit a 250K connection 😂
We run in codepipeline/codebuild
e
"250K connection" ?
z
As in, top download speed
Pipeline times out because packets being dropped etc
e
Oh right, I remember you mentioned that
Or someone else - specifically related to vendoring
z
Yeah we started caching providers in s3 but it still reaches out to the registry
And that times out sometimes for fun
So we need to set up the whole mirror
e
Ya atmos also supports mirror
Ultimately we are building up to where we can bundle everything for reproducible deployments