Hide Forgot
Description of problem: As a consequence of converting cluster.conf to be used by pacemaker, it is imperative that rgmanager be disabled. The risk of not doing this is high; it means that, after a successful conversion, it would be possible to accidentally (or intentionally) start rgmanager *and* pacemaker on the same host, managing the same resources. Consequently, a feature is proposed which has the following requirements: * rgmanager must refuse to start if it sees a particular flag in cluster.conf. * rgmanager must stop any resources and exit if this particular flag appears in cluster.conf during a reconfiguration. The proposed feature flag (in cluster.conf) is: <rm disabled="1" ... /> This requires a tiny update to the cluster schema as well.
After adding 'disabled="1"' to cluster.conf, the following occurred: node1: Aug 1 16:23:39 snap modcluster: Updating cluster.conf Aug 1 16:23:41 snap corosync[4620]: [QUORUM] Members[2]: 1 2 Aug 1 16:23:41 snap rgmanager[12407]: Resource Group Manager Disabled Aug 1 16:23:41 snap rgmanager[12407]: Shutting down Aug 1 16:23:41 snap rgmanager[12407]: Stopping service service:1 Aug 1 16:23:41 snap rgmanager[12407]: Member 2 shutting down Aug 1 16:23:41 snap rgmanager[12407]: Stopping service service:dang Aug 1 16:23:41 snap rgmanager[14427]: [script] Executing /root/uhoh stop Aug 1 16:23:42 snap rgmanager[14438]: [ip] Removing IPv4 address 192.168.122.201/24 from eth0 Aug 1 16:23:42 snap rgmanager[12407]: Service service:dang is stopped Aug 1 16:23:52 snap rgmanager[14502]: [nfsclient] Removing export: *:/mnt/gfs2 Aug 1 16:23:53 snap rgmanager[14596]: [fs] unmounting /mnt/gfs2 Aug 1 16:23:53 snap rgmanager[12407]: Service service:1 is stopped Aug 1 16:23:54 snap rgmanager[12407]: Disconnecting from CMAN Aug 1 16:24:09 snap rgmanager[12407]: Exiting node2: Aug 1 16:23:39 crackle modcluster: Updating cluster.conf Aug 1 16:23:40 crackle corosync[1918]: [QUORUM] Members[2]: 1 2 Aug 1 16:23:41 crackle rgmanager[2274]: Resource Group Manager Disabled Aug 1 16:23:41 crackle rgmanager[2274]: Shutting down Aug 1 16:23:41 crackle rgmanager[2274]: Disconnecting from CMAN Aug 1 16:23:41 crackle rgmanager[2274]: Exiting When I tried to restart rgmanager: [root@snap ~]# service rgmanager start Starting Cluster Service Manager: rgmanager disabled in configuration; not starting [FAILED] [root@snap ~]# rpm -q rgmanager rgmanager-3.0.12.1-3.el6.x86_64
Starting rgmanager after starting cman (but before quorum was attained) had the same behavior (error while running initscript) Starting rgmanager before starting cman netted with rgmanager dead, but no messages. (However, this is a "don't care" case.)
http://git.fedorahosted.org/git?p=cluster.git;a=commit;h=edc20419a06e9c96cabdf070ba676562ad22ec2c http://git.fedorahosted.org/git?p=cluster.git;a=commit;h=8e6660824136681b73506d49776f70b8a372ec63
https://www.redhat.com/archives/cluster-devel/2011-August/msg00004.html https://www.redhat.com/archives/cluster-devel/2011-August/msg00005.html
Technical note added. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: It is now possible to disable rgmanager in cluster.conf by adding disabled="1" to the <rm> tag. Adding this to cluster.conf will cause rgmanager to refuse to run, and any instance of rgmanager currently running will shut down services and exit. This feature is offered as a Technology Preview to assist users wishing to run the Pacemaker Cluster Resource Manager (which is also a Technology Preview) to prevent two resource managers from running at the same time.
Technical note updated. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. Diffed Contents: @@ -1 +1,5 @@ -It is now possible to disable rgmanager in cluster.conf by adding disabled="1" to the <rm> tag. Adding this to cluster.conf will cause rgmanager to refuse to run, and any instance of rgmanager currently running will shut down services and exit. This feature is offered as a Technology Preview to assist users wishing to run the Pacemaker Cluster Resource Manager (which is also a Technology Preview) to prevent two resource managers from running at the same time.+As a consequence of converting the /etc/cluster.conf configuration file to be used by pacemaker, rgmanager must be disabled. The risk of not doing this is high; after a successful conversion, it would be possible to start rgmanager and pacemaker on the same host, managing the same resources. +Consequently, Red Hat Enterprise Linux 6.2 includes a feature (as a Technology Preview) that forces the following requirements: + + rgmanager must refuse to start if it sees the <rm disabled="1"> flag in /etc/cluster.conf. + rgmanager must stop any resources and exit if the <rm disabled="1"> flag appears in /etc/cluster.conf during a reconfiguration.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2011-1595.html