Definitely sounds like your hitting some ceiling or have too many VMs/Containers running simultaneously. If one task (ie backup) suddenly takes more resources away from the running containers, the oom killer will cut off less ‘important’ processes in order to keep the system from crashing completely.
A completely reasonable deduction, but actually the host is swimming in headroom.
Please Login to reply.
In that case perhaps add more to the container itself? Make available an extra gig of RAM on a second core and test for a few days if that possibly alleviates the issues.
It's not a resources problem. Very low utilization.
