ASM Instance Is Hanging in an Extended Cluster When the Third Voting Disk on NFS Is Unavailable (Doc ID 1551786.1)

Last updated on AUGUST 05, 2021

Applies to:

Symptoms

Clusterware files (voting files and OCR) are stored in an ASM diskgroup.
The diskgroup used for this purpose was created with normal redundancy and contains three disks for three voting files:

* Two voting disks are located on the two SAN's
* The third voting disk is located at a third location as NFS file (e.g. on a separate Linux server)

An example for such a diskgroup (here from v$asm_disk):

Problem:

The ASM instance is experiencing a hang situation.
Dependent on the status of the asm instance and/or on the tasks which will be performed on ASM side there are different scenarios (symptoms) possible.

Some example scenarios are:

* A query from v$asm_disk or v$asm_diskgroup views is hanging.
   But at the same time a query against the v$asm_disk(group)_stat views does succeed (Note: we don't access the NFS disk physically in that case).

* An ASM diskgroup recovery is stucking. In this case we see messages in the asm alertfile like:

   14:12:51+: NOTE: waiting for instance recovery of group 3
   14:12:51+: NOTE: waiting for instance recovery of group 2
   14:12:51+: NOTE: waiting for instance recovery of group 3
   14:12:52+: NOTE: waiting for instance recovery of group 2
   14:12:52+: NOTE: waiting for instance recovery of group 1...

* The ASM instance and/or the clusterware are hanging after a restart of the CRS stack and/or of the ASM instance

Changes

Clusterware files (incl. voting files) were moved from non-ASM to ASM storage.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.