TOPIC:

I've setup Test Environment with 2 XS 6.2 Hosts iscsi-ha and HA-lizard.
Basically everything is working fine. VM get started restarted, moved between hosts etc.

However if for example both nodes die in the same time and only 1 starts which was slave before the crash then pool stays off until old master emerge.

Is this expected behavior or do I need to tweek settings ?

Please note that in my configuration switching from slave to master on master crash works perfect.

The scenario you describe works fine on our development platform running the latest releases (ha-lizard 1.7.6 and iscsi-ha 1.3.7). It should work for you. Here are some things to check:

- make sure DRBD does not auto start on boot, otherwise it will wait for its peer to appear. While waiting, the boot sequence (starting of services) stalls which could prevent ha-lizard services from starting. You can check this from the shell with "chkconfig --list" look for DRBD and ensure that all run levels are set to "off". If not, issue the following command from the shell "chkconfig drbd off"

- Make sure that you have the following set per the below, otherwise a slave cannot takeover for a Master.
FENCE_ENABLED=1
FENCE_HEURISTICS_IPS="a known and reachable IP address"
FENCE_METHOD=POOL
FENCE_MIN_HOSTS=2
FENCE_QUORUM_REQUIRED=1
FENCE_USE_IP_HEURISTICS=1

If you still have trouble with your test, try empyting the log on the slave before shutting it down (from the shell "> /var/log/messages" ) and then stop/start the slave per your test scenario. You can then post the output of the log here. There may be something preventing your slave from fencing the master. You can filter the log file with "cat /var/log/messages | grep ha-lizard"