There are three nodes which run openldap 2.0 series and are loadbalanced
trough the LVS via DR and sched is wlc, not persistant. I have noticed
that
when going trough the loadbalancer the nodes do not always reply, while if
issuing requests directly, i have a reply all the time. LDAP is a simple
tcp service, so this behavoir surprises me. I am running :
# ipvsadm -v
ipvsadm v1.20 2001/11/04 (compiled with popt and IPVS v1.0.7)
The service is pretty busy (about 1k connections at any given moment)
All your clues/solutions/requests for more info are welcome (obviously).
--
Yay! I've got a flying machine! (UNIX)
|