Cluster resources does not start post update to RHEL 8 or RHEL 9, as unfence operation fails with `fence_scsi` when configured with dm-multipath devices

Solution Verified - Updated -

Issue

  • Post updating the cluster setup to RHEL 8.10, the cluster resources does not start due to unfence operation failures, when the cluster is configured with fence_scsi agent and multipath devices are being used as the shared devices.

  • Post the upgrade to RHEL 8.10 along with all the packages, the unfence operation results in a failure similar to below:

Aug 31 04:18:09 node-ha1 /fence_scsi[301305]: Cannot get registration keys on device "/dev/disk/by-id/scsi-3600000000000000000000000000": Persistent reservation in:#012Fixed format, current; Sense key: Unit Attention#012Additional sense: Registrations preempted#012PR in (Read keys): Unit attention#012sg_persist failed: Unit attention
Aug 31 04:18:09 node-ha1 /fence_scsi[301305]: Please use '-h' for usage
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ 2024-08-31 04:18:09,897 ERROR: Cannot get registration keys on device "/dev/disk/by-id/scsi-3600000000000000000000000000": Persistent reservation in: ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ Fixed format, current; Sense key: Unit Attention ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ Additional sense: Registrations preempted ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ PR in (Read keys): Unit attention ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ sg_persist failed: Unit attention ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ 2024-08-31 04:18:09,897 ERROR: Please use '-h' for usage ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: warning: fence_scsi[301305] stderr: [ ]
Aug 31 04:18:09 node-ha1 pacemaker-fenced[256524]: error: Operation 'on' [301305] targeting node-ha1 using scsi returned 1 
  • With upgrading the fence-agents-scsi version to fence-agents-scsi-4.2.1-129.el8 or later, the unfence operation results in a failure although the kernel version upgrade to RHEL 8.10 was performed much before the update of fence-agents-scsi package.

Environment

  • Red Hat Enterprise Linux 8, 9
  • High Availability Pacemaker Cluster
  • fence-agents-scsi-4.2.1-129.el8 or later
  • Cluster managing shared storage over multipath with fence_scsi agent configured

Subscriber exclusive content

A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.

Current Customers and Partners

Log in for full access

Log In

New to Red Hat?

Learn more about Red Hat subscriptions

Using a Red Hat product through a public cloud?

How to access this content