Alvaro Aleman

@danderson So your issue is only about being notified about a synced cache before any reconciliation started? If I understand the issue correctly, this also implies that this "Init code"...

Yeah, this is a common-ish problem and something that controller-runtime doesn't help with. On a conceptional level, you likely want the handler itself to be a controller (i.E. have a...

/cc @chaodaiG Maybe it would be best to plug `validateStoragePath` into the `io.Opener` as that is the common code that is used for all object storage access?

Yeah, the config is pretty much moved out of the prow directory. What would be needed to move prow into its own repo is: * The repo itself * Exporting...

Doesn't storage.create mean you can "delete" content by writing a new empty file to an existing location?

Yeah I saw this in Slack but I doubt I will have time to look into this in the near future. Some of the screenshots @ybettan showed in Slack also...

The issue is that that the event (a timer) that causes us to load the job config and then trigger the job is part of said config itself. In the...

> scrape every time horologium runs This isn't doable, because horologium runs every 30 seconds and would need to fetch * All branches of all repos * Check if the...

> What's in my mind was a separate goroutine doing this update loop, which of course will cause delay, but I guess it's bearable if it's a couple of minutes...

> Yeah if we want to be more correct and more resilient in general then this seems like the approach to go, might run into configmap 1Mb size limit though....