rgmanager does not resume managing resources after losing then regaining quorum when using central_processing mode in a RHEL 5 or 6 High Availability cluster
Issue
rgmanager
stop working when quorum is dissolved on the surviving cluster node on RHEL 6.- After the cluster lost quorum, the clustered services were not running on the surviving cluster node.
clustat
output does not show the clustered services after quorum is dissolved and regained when using qdisk.- We had an issue in our cluster that caused nodes to lose quorum, but after that problem was fixed,
clustat
isn't reporting any service states or Master/Worker states for any nodes. clusvcadm
returns "Service does not exist" for a service that definitely exists in the configuration
# clusvcadm -e nfsexport
Local machine trying to enable service:nfsexport...Service does not exist
rgmanager
seems to stop working after we have a membership event and loss of quorum- After rebooting a cluster node the clustat output on that node shows the services in an "uninitialized" state:
# clustat
Cluster Status for myCluster @ Mon Feb 3 16:37:44 2014
Member Status: Quorate
Member Name ID Status
------ ---- ---- ------
node1 1 Online, rgmanager
node2 2 Online, rgmanager
node3 3 Online, rgmanager
node4 4 Online, Local, rgmanager
Service Name Owner (Last) State
------- ---- ----- ------ -----
service:service1 none uninitialized
- When starting a service using 'clusvcadm -e' we are getting 'is now running on (null)' message.
# clusvcadm -e my_service
Local machine trying to enable service:my_service...Success
service:my_service is now running on (null)
Environment
- Red Hat Enterprise Linux (RHEL) 6 with the High Availability Add-On
<rm central_processing="1"/>
in/etc/cluster/cluster.conf
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.