So i have two nodes, 1-1 NIC for DRBD replication, ietd to connect the storage and heartbeat to keep them in sync.
Everything works fine, a tested it with restarting heartbeat as written in the last point. i was watching /proc/net/ietd/session, and the change is OK, the session builds up very quick.
The problem comes when i restart the heartbeat on active node first time after boot. I have to wait for ietd for about 20-30 sec to send ISCSI logins.in this period the sessions build up, and the connection works again but very slow. After this first change will the change fast again untill the next reboot. I was looking for differences in logs but found nothing. Do you have any idea why the session build up so slow at the first heartbeat restart?
Syslogs, ha-debug logs, ISCSI shark and conf's are here: