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.