Cluster lock disk /dev/dsk/c4t0d0 has failed: I/O error

[Rambo_1]

Hello

I have a cluster of 2 machines. The packages use disks in the va7100 arrays. I get this messages as below:
cmcld: WARNING: Cluster lock disk /dev/dsk/c4t0d0 has failed: I/O error
cmcld: Until it is fixed, a single failure could
cmcld: cause all nodes in the cluster to crash

there is only one lun in va7100, I mean that the lun 0 is the c4t0d0 .how to solve this problem ？

Thanks a lot for your support

[Henk Geurts]

hi Rambo
looks like a hardware problem....

did you check the disk is accessable from both nodes in the cluster?

do an
ioscan -fnC disk
to see if it's CLAIMED.
also
do a vgdisplay -v
to see if the disk is "available".

let me know.
regards.

[Mark Nieuwboer]

Hi Rambo,

As you probebly know by now you have a harware problem. What doen mail root say.
Futher the trouble is when you have to stop and start your cluster it won't come up because he need the disk to start the cluster. If one nodes in the cluster can't lock the disk the nodes things the other machine has the disk and waits for the cluster to reform. So nothing happens.
fix your hardware problem most like replace the disk.

grtz. Mark

[melvyn burnard]

So if there is only one LUN for the cluster to use, can it be seen ok by both nodes?
Are both nodes logging the same WARNING message (it will be every 60 minutes as that is the polling interval for the CL disk)?
Is the package running ok?
If the answer to the above questions are all yes, then it appears that something has happened recently to remove the setting on the disk that it is used for cluster locking.
You have two options here.
1) Halt the cluster, vgchange -c n vg01, vgchange -a y vg01, cmapplyconf -C clusterasciifile, vgchange -a n vg01, cmruncl and then check to see if the messages re-appear.
2) log a call with your local HP Response Centre and request the cminitlock tool.

If not, then you have a problem on your disk array, and you may need to get it checked out.

Your cluster should still be able to stop and start without the cluster lock disk, as it only needs the cluster lock at the time of a cluster reformation that results in EXACTLY 50% of nodes being available.

[Mohanasundaram_1]

Hi Rambo,

confirm if the LUN in question is accessible by your systems at present. If they are, then just perform

# ioscan -fn

If you then see a message in syslog.log saying "the cluster disk had come online again" you need not worry.

I am presuming here that the LUN had a momentary problem and has come back now.

I have seen this happen in my environment once.

With regards,
Mohan.

[Rambo_1]

Hi, all
Thanks a lot for your help .
Hi melvyn burnard,I will take a time to check&test the cluster, and then I will give the feedback&point to you. , thanks for your support again.

BR