Blog Entry

fdeckert's picture
blog
Reads:

5184

Score:
0
0
 
Comments:

1

Tribulation of an OES admin: iSCSI OES linux cluster and missing SBD partition

Author Info

17 January 2008 - 10:44am
Submitted by: fdeckert

(View Disclaimer)

We migrated cluster pools from an old iSCSI storage to a new one, and of course we forgot to migrate the SBD partition.
At boot time, the nodes were complaining they found no SBD partition, and naturally we got no NCS.
After boot, no way of creating a new SBD partition or access partitions list thru yast. sbdutil or yast get stuck

This was tricky to figure out and fix. Let me resume:
-we copied data from on iSCSI storage (nw65) to a new one (sles10)
-we forgot to copy the SBD partition
-of course we unplugged the nw65 storage
-and we booted the cluster nodes.

As no SBD partition was available, rcnovell-ncs complained and refused to join the cluster on both nodes.

We tried to create a SBD partition as written in the documentation : "sbdutil -c /dev/sdd". But sbdutil gets stuck and never creates the partition. Neither could we create a partition with yast or evmsmu, all stuck.

Only way to access the partitions ... boot without ncs. so a "chkconfig novell-ncs off && reboot". Yep that time we can create a SBD and we enforce the cluster name "sbdutil -c /dev/sdd -n MAGIC".

But at reboot time, with NCS back on : "no sbd partition to be found". Hey, we just created on !

After many tries and head bumps we found the final trick : one has to initialize the SBD partition after creation ! And of course find the correct partition path ;-)

So the correct way of re-recreating an SBD partition is:
- chkconfig novell-ncs off
- reboot
- sbdutil -c /dev/sdd -n CLUSTERNAME
- sbdutil -i -p /dev/evms/.nodes/clustername.sbd -n CLUSTERNAME
- chkconfig novell-ncs on
- reboot

Hum, maybe I shall propose a TID on the subject :-)


Disclaimer: As with everything else at Cool Solutions, this content is definitely not supported by Novell (so don't even think of calling Support if you try something and it blows up).

It was contributed by a community member and is published "as is." It seems to have worked for at least one person, and might work for you. But please be sure to test, test, test before you do anything drastic with it.




User Comments

wolfout's picture

missing option

Submitted by wolfout on 9 July 2009 - 10:31am.

Your instructions were very good EXCEPT in this line:
sbdutil -c /dev/sdd -n CLUSTERNAME
You need to have it look like this:
sbdutil -c -d /dev/sdd -n CLUSTERNAME

© 2013 Novell