I got OOM on both the secondary (non-ca) and 2 days after on the master. It seem it was “icinga2” which ate all of the memory and was killed in the OOM. After that the host regained the memory and looked normal.
Both nodes are the same OS kernel version CentOS 7.9
IcingaWeb2 - 2.7.3
Enabled features: api checker ido-mysql mainlog notification
Config validation (
icinga2 daemon -C) - with some warning it goes through okay
Turned out the maridab was hosed due to running out of disk space.
I think icinga2 should eat all memory in such scenario.