Briefing
The api Deployment moved to the new cluster ten minutes ago with a manifest the platform team tidied up on the way.
Since then all three pods have been restarting every minute or two, and the Deployment has not shown a single available replica.
The image, myapp:v1, is the same build that ran for weeks on the old hosts, and its logs look like a normal boot every time.
Dev is convinced the new nodes are short on memory.
The fix is yours to find: hints, the recovery checks and the debrief unlock inside the incident.