[Spread-users] Spread errors using spreadlogd...

[ken stumpf]

I'm using spreadlogd in a cluster of seven machines.  They are high traffic servers for a site that pushes about 120mbit a day.  There is a spread daemon on each machine, communicating via multicast.  The spread daemon seems CPU hungry, using about 20% of cputime.  Is this normal?  The apache error logs are getting bombarded with this message:

[Wed Oct 18 01:15:48 2006] [notice] [client 66.249.66.17] Sending message to spread daemon 0 with group wwwgroup failed. During communication to send the message errors occured and the send could not be completed. Trying to reconnect, attempt 1.

I couldnt find any previous mention of this error, so I'm not sure whats happening or how to tweak spread to get rid of it.  I've tried running with MAX_SESSION_MESSAGES set to 10000 and 20000 to no avail.  Ive read the spread users guide but I do not see any information there to further tweak my setup.  Can anyone point me in the right direction?

Thanks!
Ken