We have an 11gR2 RAC (GI) cluster, it was upgraded from 11gR1 (CRS). We have to move our Voting disk off of old storage and onto new. Our current voting disk is not in ASM, but we are trying to migrate it to ASM. A diskgroup was created with Normal Redundancy and with ASM Compatibility at 11.2. We migrated OCR successfully to the new ASM diskgroup, confirming with ocrcheck on all nodes of the cluster. Next we ran the command on one node to replace voting disk which failed: crsctl replace votedisk +OCR_1
Here is the output:
Error while updating the profile with ASM discovery string
CRS-4000: Command Replace failed, or completed with errors.
Looked at ASM alert.log and found the following.
CRS-2324:Errors occurred while trying to push GPnP profile.
A trace file shows the following, repeated several times:
2013-12-05 15:06:49.414: [ GPNP]clsgpnp_findSrv2: [at clsgpnp0.c:1835] FIND INCOMPLETE: Not all requested 'gpnp' services on 4 host(s) were found/resolved - discovery retries (0 of 6) will continue after short sleep.
I've run through the following commands to troubleshoot and all looks fine.
$GRID_HOME/bin/crsctl check cluster -all
$GRID_HOME/bin/cluvfy comp gpnp -verbose
$GRID_HOME/bin/oifcfg getif -global
$GRID_HOME/bin/oifcfg iflist -p
I reviewed the GPnP profile.xml using $GRID_HOME/bin/gpnptool get
Here is an excerpt showing the voting disk on the old storage instead of in ASM diskgroup
<orcl:CSS-Profile id="css" DiscoveryString="/dev/emcpowerb1" LeaseDuration="400"/>
I looked through many online docs and forums trying to find out why the profile is not updating but am at a deadend and asking for help. If I can figure out why the profile isn't updating, I may be able to figure out a solution.
Yes, thanks. I had already confirmed that 11.2 compatiliby was set. It was run as root. But I figured it out! Though the gpnptool commands I had tried running to troubleshoot came back ok, when I ran "gpnptool find", the result showed instances of gpnp for all our RAC nodes except 1. I then killed the gpnpd.bin process on that node, crs restarted it immediately. Now "gpnptool find" shows all nodes of the cluster. I was able to run the crsctl replace votedisk command successfully.