Redhat Cluster (Pacemaker/Corosync): DLM Not Starting

Question

I need help regarding my cluster error:

[root@db2]# pcs status
Cluster name: oracluster
Last updated: Mon Feb 22 16:00:12 2016
Last change: Mon Feb 22 15:45:14 2016
Stack: corosync
Current DC: db2 (2) - partition with quorum
Version: 1.1.12-a14efad
2 Nodes configured
5 Resources configured


Online: [ db1 db2 ]

Full list of resources:

 ClusterVIP (ocf::heartbeat:IPaddr2):   Started db2
 Clone Set: dlm-clone [dlm]
     Stopped: [ db1 db2 ]
 Clone Set: clvmd-clone [clvmd]
     Stopped: [ db1 db2 ]

Failed actions:
    dlm_start_0 on db2 'not configured' (6): call=18, status=complete, exit-reason='none', last-rc-change='Mon Feb 22 15:57:04 2016', queued=0ms, exec=34ms


PCSD Status:
  db1: Online
  db2: Online

Daemon Status:
  corosync: active/disabled
  pacemaker: active/disabled
  pcsd: active/enabled

Details:

I have 2 nodes (db1, db2) with shared storage (SAN). Both servers are in RHEL7.1. Now I want to add the storage as a resource. According to RHEL documentation DLM and CLVMD should be added also as a resource. I discovered that the error will disapper when STONITH is enabled, but still DLM is not starting. The log says it needs Fencing Device to be configured, which I don't have right now.

Any work around for this? Do we have a way to disable the fence mechanism and still make the cluster work? Thank you so much in advance!

Answer 1

you said that you have san storage, then you create a partion for fencing and use it as scsi stonith, il will solve your problem, like this exemple:

pcs stonith create scsi-stonith-device fence_scsi devices=/dev/mapper/fence pcmk_monitor_action=metadata pcmk_reboot_action=off pcmk_host_list="node1 node2" meta provides=unfencing

and don't forget to enable stonith with pcs property set stonith-enabled=true

Answer 2

Configure SONITH. It will help you to fix this issue.