Image Automation controller silently stops working

[sjdweb]

Describe the bug

On our clusters, we had connectivity issues on Dec 26th. I noticed that since then, image automation failed to update (but ImagePolicies were up to date).

Here are the last logs for one of the controllers:

{"level":"error","ts":"2021-12-26T21:37:24.895Z","logger":"controller-runtime.manager.controller.imageupdateautomation","msg":"Reconciler error","reconciler group":"image.toolkit.fluxcd.io","reconciler kind":"ImageUpdateAutomation","name":"flux-system","namespace":"flux-system","error":"unable to clone 'ssh://[email protected]/myco/fleet-infra', error: SSH could not read data: Error waiting on socket"}

Since that point in time, the controller stopped working.
Killing the pod fixed this issue, but it'd be great if it could self-heal in this scenario?

Steps to reproduce

1. Allow image automation controller to have connectivity issue from Git repo
2. Observe that the controller will not try again to connect, or crash

Expected behavior

The controller should try again to reconcile as the connectivity would have been resolved.

Screenshots and recordings

No response

OS / Distro

Ubuntu 20.04

Flux version

flux version 0.16.1

Flux check

N/A

Git provider

GitHub

Container Registry provider

No response

Additional context

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[aholbreich]

Same for us.

but it alls seem to be go down to git level if i not mistaken. fluxcd/source-controller#439 (comment)

[squaremo]

@aholbreich There are two related problems:

1. getting "error: SSH could not read data: Error waiting on socket" in the logs
2. the image-automation-controller stops doing anything after that message

I know you are seeing the log message, from the comment you linked -- are you also experiencing the second problem, that the controller stops doing anything?

[aholbreich]

Hi, no. Not really seen controller problems.

[pjbgf]

Same as #282. I will be updating that thread instead, here's the latest comment:

#282 (comment)