Service guard cmquery question

[Allen DuBeau]

When I ececute cmquerycl -v on of the nodes in the cluster is listed as "out of date"

Does this mean the cluster configuration file is out of date or that service guard version is out of date?

When I execute cmgetconf -c temp.ascii

I get several LVM error messages, but no ascii file generated:

Volume group /dev/vgcluster is configured differently on node NODE1 than on node NODE2

Error: Volume group /dev/vgdata on node NODE2 does not appear to have a physical volume corresponding to /dev/dsk/c13t4d3 on node NODE1(1617719311110321321).


Thanks...

Allen

[Geoff Wild · Accepted solution]

That's not good.

You need to reapply your cluster config file.

Is the cmgetconf failing on all nodes or just 1?

If only one, then on the "good" node - run:

cmgetconf -v temp.ascii

Then do a cmcheckconf -v -C temp.ascii

If okay - then do a cmapplyconf....

You may have to vgexport the /dev/vgdata on the NODE2, then re-import that vg.

Rgds....Geoff

[Kent Ostby]

So these messages are telling you that the volume group definitions vary between machines.

If you are sure about which disks are suppossed to be where, I would do the following.

On the working node do -p vpexport to create a map file:

vgexport -p -s -m /tmp/map.vgdata /dev/vgdata

then copy it to node 2

On node 2, run vgexport to get rid of the lvmtab data;

vgexport /dev/vgdata

recreate the datastructures on Node 2:
mkdir /dev/vgdata
mknod /dev/vgdata/group c 64 0xhh0000

NOTE: the 0xhh0000 number should match whatever is on node 1.

Finally do the vgimport:

vgimport -s -m /tmp/map.vg07 /dev/vg07