-
Problem report
-
Resolution: Unresolved
-
Trivial
-
None
-
6.0.16
-
None
-
Ubuntu 20.04
Steps to reproduce:
- Have ServerActive with three servers in an HA cluster
- Have the leader stop
Result:
Zabbix_agent2 keeps trying to connect to the leader once a second and get a RST packet back:
$ sudo tcpdump -n 'tcp[13] & (tcp-rst) != 0' 04:35:45.011937 IP 11.11.11.11.10051 > 22.22.22.22.35389: Flags [R.], seq 0, ack 3810576045, win 0, length 0 04:35:46.018579 IP 11.11.11.11.10051 > 22.22.22.22.42077: Flags [R.], seq 0, ack 2494288133, win 0, length 0 04:35:47.024100 IP 11.11.11.11.10051 > 22.22.22.22.55065: Flags [R.], seq 0, ack 2133648351, win 0, length 0 04:35:48.031102 IP 11.11.11.11.10051 > 22.22.22.22.52909: Flags [R.], seq 0, ack 2917491057, win 0, length 0 04:35:49.037545 IP 11.11.11.11.10051 > 22.22.22.22.57101: Flags [R.], seq 0, ack 2796766943, win 0, length 0 04:35:50.004574 IP 11.11.11.11.10051 > 22.22.22.22.54617: Flags [R.], seq 0, ack 1171054559, win 0, length 0 04:35:51.011327 IP 11.11.11.11.10051 > 22.22.22.22.37697: Flags [R.], seq 0, ack 1585614916, win 0, length 0
Restarting zabbix_agent2 causes proper operation again.
Expected:{}{}
Zabbix should stop pinging the dead leader every second after a minute, and revert to once-a-minute pings or longer.