The default JGroups openshift.DNS_PING discovery protocol log level is verbose way too much. Especially when scaling pods down.
To avoid the pollution of the log of the pod, when scaling cluster down, the log level for JGroups openshift.DNS_PING should be user configurable:
- Maybe default to log FATAL messages only (not to show that huge amount of errors about error to connecting to pods, that have been terminated),
- And if JGROUPS_PING_LOG_LEVEL variable is set, derive the log level for the JGroups openshift.DNS_PING protocol from the value of that variable.
- Per upstream documentation, JGroups supports setLevel() routine,
- Which transfers into level attribute of the <protocol> XML definition per this question
CLOUD-2236 [EAP] Verify clean pod termination on scale down for tx recovery