On Mon, Feb 23, 2004 at 01:41:36PM -0500, Steve Shepherd wrote: > We are running 18 pop servers in an LVS/HA configuration, some of our > clients complain of an almost "immediate" timeout (within 5-10 seconds) > of logging into their mailbox with sqwebmail. All servers set exactly > the same: > > SetEnv SQWEBMAIL_TIMEOUTSOFT 43200 > SetEnv SQWEBMAIL_TIMEOUTHARD 43200 > > I am unable to replicate this issue by going to each individual server. > Any ideas?
You modified cleancache.pl for the new timeout? Your load-balancer applies a "sticky" algorithm so that all hits from the same source IP address hit the same one of the 18 servers? If so, I don't know, sorry... it seems to work OK for me (but with only 2 webmail servers plus 3 pop+imap servers with an NFS backend) Brian.
