label bot workers stop receiving pubsub messages; issue with workload identity?

[jlewi]

trafficstars

As part of #70 we deployed the workers on an update cluster which uses workload identity.

I'm observing that after the workers have been up for a long time they appear to stop receiving pubsub notifications.

This is visible in cloud console as a growing backlog of messages.

I suspect an issue related to credentials and workload identity. Bouncing the pods appears to fix it.

Related to #70

[issue-label-bot[bot]]

Issue-Label Bot is automatically applying the label kind/bug to this issue, with a confidence of 0.96. Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback!

Links: app homepage, dashboard and code for this bot.

[kf-label-bot-dev[bot]]

Issue-Label Bot is automatically applying the labels:

Label	Probability
kind/bug	0.96

Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback! Links: app homepage, dashboard and code for this bot.

[jlewi]

To try to recover

• Delete the gke metadata servers

kubectl -n kube-system  delete pods -l k8s-app=gke-metadata-server

• Restart the label bot pods

kubectl delete pods -l app=label-bot