patch: panic if unable to get ecr creds

[c-fteixeira]

Even though we could track/parse logs, a failing container seems more appropriate if it is unable to do its main task. Let me know if this is too drastic..

[gflarity]

Now that we're using spiffe-jwt instead, there were still issues? It does seem a bit drastic but if something is happening regularily this might be needed.

I believe this will only restart the container though and not the whole pod, so if there's a bug/issue with spiffe-jwt, panic'ing like this might not help at all.

[c-fteixeira]

there were still issues?

I would not call it "regularly", but yes, i've noticed a few instances.
The restarts would only make it easier to detect, even if it doesn't auto fix for now.

[gflarity]

I guess we have no idea why it's happening?

[gflarity]

Anyways, this is fine for now. I wonder if just having metrics around failures with improve the alerting and then make it easier to debug while it's happening. I've created a ticket for integrating metrics.