cmcheckconf core dump

[Michael O'brien_1]

Hi,

I working on a RAC cluster and have had a requirement to add a new volume groups to the running cluster. So I ran a cmgetconf for the cluster, than when I ran a cmcheckconf using the ascii file produced by the cmgetconf. Several messages where displayed, saying unable to query /dev/vg00 on several of the nodes and then query time-out. This was followed by a core dump. I plan of sending the core dump to HP for analysis but I was just wondering if any one could shed any light on why it's querying the /dev/vg00 disks.

Thanks
Michael

[Stephen Doud]

cmquerycl has to determine which VGs are shared between nodes. The code doesn't distinquish VG names (in /etc/lvmtab), so it inspects the VGID on every VG listed. When it finds a common VG name AND VGID, it found a shared VG and will investigate the integrity of that VG further, as part of the cmcheckconf process to validate the cluster.

As for the core dump - suggest you cksum the cmcheckconf command on each node to see if it is the same between the two.