flux will not recover from failed deployments

[alita-moore]

I might be doing something wrong but when a deployment fails (i.e. there is an ImagePullBackoff error or a config error) flux fails to re-deploy the deployment so I have to manually run kubectl rollout deployment whatever for the reconciliation to get unstuck. how can I make flux do this automatically?

[stefanprodan]

Flux will redeploy it after the timeout expires, you should also set the retry interval, see here the recommended settings: https://fluxcd.io/flux/components/kustomize/kustomizations/#recommended-settings

[alita-moore]

it seems that even with these settings it's not recovering / retrying deployments for knative services

[alita-moore]

i.e. it would be potentially a good idea to force a new revision of the knative service in the event of a timeout / failed deployment. Doing so manually resolves the problem. (I do so by setting some annotation on the knative service).