Cluster fails to fence a node with fence_scsi and logs report "Failed: Cannot open file '/var/run/cluster/fence_scsi.key'" in a RHEL 7 Update 1 High Availability cluster
Issue
- If I reboot a node, after it rejoins then if it needs to fence another node it fails with an error in
/var/log/messages
saying it couldn't open/var/run/cluster/fence_scsi.key
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: error: log_operation: Operation 'reboot' [3074] (call 2 from stonith_admin.3047) for host 'rhel7-node2.example.com' with device 'scsi' returned: -201 (Generic Pacemaker error)
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ WARNING:root:Parse error: Ignoring unknown option 'port=rhel7-node2.example.com' ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ERROR:root:Failed: Cannot open file "/var/run/cluster/fence_scsi.key" ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ Failed: Cannot open file "/var/run/cluster/fence_scsi.key" ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ERROR:root:Please use '-h' for usage ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ Please use '-h' for usage ]
Sep 15 14:22:08 rhel7-node1 stonith-ng[2950]: warning: log_operation: scsi:3074 [ ]
Sep 15 14:22:09 rhel7-node1 stonith-ng[2950]: notice: stonith_choose_peer: Couldn't find anyone to fence rhel7-node2.example.com with <any>
Sep 15 14:22:09 rhel7-node1 stonith-ng[2950]: error: remote_op_done: Operation reboot of rhel7-node2.example.com by <no-one> for stonith_admin.3047@rhel7-node1.example.com.49e1b746: No route to host
- Unfencing isn't creating files
/var/run/cluster/fence_scsi.key
and/var/run/cluster/fence_scsi.dev
- My cluster can't fence
- I'm using
fence_scsi
andpcs stonith fence
returnsno route to host
, but only after I've rebooted a node and rejoined it to the cluster
# pcs stonith fence rhel7-node2.example.com
Error: unable to fence 'rhel7-node2.example.com'
Command failed: No route to host
Environment
- Red Hat Enterprise Linux (RHEL) 7 with the High Availability Add On
fence-agents-scsi
releases starting with4.0.11-11.el7_.1
up to (but not including)4.0.11-13.el7_1.2
- One or more
stonith
devices configured to usefence_scsi
- If using a release of
fence-agents-scsi
that is susceptible to monitor failures, then the issue described in this solution is only applicable if thestonith
device is configured with the appropriate workaround or resolution.
- If using a release of
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.