NodeLeave gossip event reported when cluster under heavy load

[tgruben]

Expected behavior

only report node leave when either node crashes or signals a departure from the cluster

Actual behavior

when cluster performing a long running operation > hours. gossip is determining that a node has died du to lack of response to probes.

Information about your environment (OS/architecture, CPU, RAM, cluster/solo, configuration, etc.)

3 node oci cluster 930 GB ram 144 cores 72 TB nvme storage

[jaffee]

Possible that increasing gossip.suspicion-mult will improve this behavior. Need to test.