Can you confirm that pacemaker's stonith is working itself? If you
temporarily remove DRBD from pacemaker, stop the DRBD service and then
try crashing a node, does it get rebooted? I do this test by calling
'echo c > /proc/sysrq-trigger', which causes the node to immediately panic.
The 'out of memory' issue is not something I've ever seen before. Once
we confirm that pacemaker's fencing is working, we can look into it.
What version of pacemaker, corosync and drbd are you using?
digimer
PS - If fencing fails, DRBD _should_ remain blocked, so that behaviour
is expected and good.
On 02/04/14 02:58 PM, khaled atteya wrote:
Dear Digmer,
drbd conf : http://fpaste.org/91031/45805613/
pacemaker conf : http://fpaste.org/91036/13964584/
Now , when i drop the DRBD link in n2.stg node , these logs appear :
http://fpaste.org/91040/45990213/
note "out of mem, failed to invoke fence-peer helper"
Fencing didn't happen , but I/O is blocked .
Where may the problem ?
On 4/1/14, Digimer <[email protected]> wrote:
On 31/03/14 12:26 PM, khaled atteya wrote:
Hi
I installed cluster (corosync+pacemaker+drbd) , and i am using IPMI as a
fencing device .
I have 2 link , one of them for drbd , and pacemaker use both links .
When I cut the 2 links (by drop connection by iptables) , fencing is
working good , but when i cut the drbd link (also by iptables) nothing
fencing happen .
why?
Do you have DRBD configured in pacemaker and did you set the
fence-handler to be /path/to/crm-fence-peer.sh?
If this doesn't resolve the issue, please shared you 'drbdadm dump' and
pacemaker configurations.
--
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?
--
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?
_______________________________________________
drbd-user mailing list
[email protected]
http://lists.linbit.com/mailman/listinfo/drbd-user
