vxconfigd fails to accurately change disk I/O policy (4197480)

Last published : Sep 23, 2026
On a CVM cluster, when disk paths are disabled on a slave node and vxconfigd is stopped and then restarted (vxconfigd -k -m enable), vxconfigd fails to enable. GAB ports u and w may become unavailable on the node.
For every disk belonging to a shared disk group, the following types of messages appear in syslog:
VxVM vxio V-5-3-2493 new_disk: <disk_name> exists
VxVM vxio V-5-3-2473 disk_change_iopolicy: DG ID mismatch 0.0 : <dg_id_hi>.<dg_id_lo>
Additionally, running the vxdctl enable command results in the following error:
VxVM vxdctl ERROR V-5-1-1589 enable failed: Error in disk group configuration copies
This issue occurs due to the following sequence of events. When vxconfigd is killed and restarted on a CVM slave node, vxconfigd attempts to reload each disk and change its I/O policy. As the disks already exist, the disk reload is rejected, and the V-5-3-2493 error code is logged. However, vxconfigd still proceeds to change the I/O policy. It also incorrectly sends a disk group ID of 0.0, instead of the actual disk group ID, which leads to disk group ID mismatch.
Workaround:
  1. Re-enable all the disabled disk paths.
    `# vxdmpadm enable path=<disk_path>```
  2. Restart the vxconfigd program.
    # vxdctl enable
  3. (Optional) If the GAB ports are down, run the following commands sequentially to attempt a recovery:
    # vxclustadm stopnode
    # vxclustadm -m gab -t gab startnode