monit<->mmonit connections

[Gerrit Kühn]

Hi all,

In some cases I see very weird behaviour of monit opening tons of
connections to mmonit that remain in status TIME_WAIT:

tcp        0      0 192.168.112.7:41005     10.117.15.111:8080      TIME_WAIT  


 ~ # netstat -an|grep -c TIME_WAIT        
8153


These connections also appear to eat up netfilters conntrack table:

[4993154.310279] nf_conntrack: table full, dropping packet.


I cannot see any reason for this happening. I see it only few times (maybe 
thrice over the last year while running monit on about 50 machines) on 
different machines. It does not go away by restarting monit. However, rebooting 
the machine seems to help.
Any idea what might cause this behaviour and how to get rid of it?


cu
  Gerrit