cman fails to start when a node reboots while an LVM physical volume is missing or dead and fence_scsi is being used in a RHEL 6 Resilient Storage cluster
Issue
- Power supply to one of two SANs is gone, causing the clustered LVM mirror to be converted to a linear device. Now if in this situation a node needs to be rebooted, that node won't be able to join the cluster because the scsi unfencing will fail.
- cluster with two SANs when one SAN is down and a node is rebooted it fails to join the cluster
- If I reboot a node while one physical volume in the clustered volume group is failed,
fence_scsifails to unfence the node andcmanwon't start fence_scsifails to unfence with a missing physical volume. Is there any automatic way to recover from this so that a node can reboot while a device is missing?
Environment
- Red Hat Enterprise Linux (RHEL) 6 with the Resilient Storage Add On
lvm2-clusterlocking_type = 3in/etc/lvm/lvm.conf- One or more volume groups with the clustered attribute set
- One or more devices in a clustered volume group is missing or dead
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.
