Description of problem: A cluster sevice involving clusterfs resources will be infinitely recovered. Version-Release number of selected component (if applicable): rgmanager-1.9.39-0 How reproducible: always Steps to Reproduce: 1.create a cluster service using a GFS filesystem 2. 3. Actual results: service is considered malfunctioning, rgmanager will continuously relocate and recover the service Expected results: rgmanager should detect that the service is not malfunctioning. Additional info: In clusterfs' status/monitor code, there is a test that has to fail if the service is OK ([ $? -ne $YES ] && exit $OCF_ERR_GENERIC). But the last line returns the last exit code (exit $?). So the status always returns 1. This mini patch fixes this behaviour: --- /usr/share/cluster/clusterfs.sh~ 2005-10-18 21:17:51.000000000 +0200 +++ /usr/share/cluster/clusterfs.sh 2005-10-30 14:57:06.000000000 +0100 @@ -846,7 +846,7 @@ isAlive ${OCF_RESKEY_mountpoint} [ $? -ne $YES ] && exit $OCF_ERR_GENERIC - exit $? + exit 0 ;; restart) stopFilesystem
Patch has been applied to the stable, rhel4, and head branches.
This is addressed, so I'm setting state to modified for now -- it should come out with the next update.