You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
So far I have not found a way to implement this. Even if we exec into the pod, there is no way to terminate Renovate without the pod being marked as "Failed", and subsequently getting restarted.
ps aux inside the pod shows the following process hierarchy:
No matter which of these processes I terminate (kill -s SIGTERM <pid> or kill -s SIGKILL <pid>), the pod is marked as failed - which makes sense, I guess.
However, this leads me to believe we can only "cancel" a run by deleting the Job itself - not the Pod. In that case, any logs would be gone, along with any trace of this action being performed. This seems bad from a monitoring/auditability perspective.
I suppose in theory we could override the command of the Renovate container so that it runs Renovate as usual, but traps SIGTERM & exits gracefully. This might work for runs triggered manually via the Foreman UI, at least. Arguably it's more likely someone wants to cancel a manual ran than a scheduled run. However, this seems like a fragile implementation at best.
No description provided.
The text was updated successfully, but these errors were encountered: