We're still looking into the root cause of the issue - There was an out of memory situation which caused containers being rescheduled on other compute nodes.
Posted Jun 08, 2021 - 10:14 UTC
Identified
The issue has been identified and a fix is being implemented.