Trigger all resource checks in a pipeline with a single fly command #9736
Replies: 2 comments 4 replies
|
I think it is a good idea! I think most operators might already have implementations for this, as it is easily scriptable, but that being said some commands support (-a, —all) flags, so not sure if it was intentional not added, or just missed. Maybe @taylorsilva can comment. What is a bit worrying to me is that big team might have a lot of resources and such parallelism might cause “max containers reached” errors, since a lot of containers would be scheduled in parallel. Did you consider that part. As checks are quite fast, usually when you click one by one, they already succeeded, so such bottleneck might be missed. Have you tested on large instances without global resources? Or at least considered some sort of max parallelism etc. |
|
Converted discussion to an issue |
Uh oh!
There was an error while loading. Please reload this page.
Hello,
Concourse already supports
fly check-resource -r <pipeline>/<resource>which works great for a single resource, but there is no way to trigger a check for all resources in a pipeline at once.After maintenance windows or infrastructure changes it is common to need to force-recheck every resource in a pipeline to make sure everything picked up the change. With a pipeline of 20 resources this means running fly check-resource 20 times manually, which is tedious and easy to miss one.
My idea is to create a new
fly check-pipeline-resources -p <pipeline-name>command that triggers a check for all resources in the pipeline in one shot. After running it I'd expect all resources to start a new check immediately, the same way fly check-resource does for a single one.I have a working POC and I'd be happy to discuss it further with the community and open a PR if there is interest. Looking forward to hearing your thoughts!
All reactions