cmgetconf (cmgetconf -c <cluster name> output file )is not creating output file..

[Jagur]

Hi,

 I need to migrate my servers  from 11.23 to 11.31 .There are 3 server are part of the cluster..As part of pre-work,I tried to collect the configuration using cmgetconf,but not creating any output file...

ieoesh09# cmgetconf -v -c ieoeshclqabi1 test.ascii
Gathering storage information
Volume group /dev/vgqawt1jd1 is configured differently on node ieoesh28 than on node ieoesh15
Volume group /dev/vgqawt1jd1 is configured differently on node ieoesh15 than on node ieoesh28
Volume group /dev/vgqawt1jd1 is configured differently on node ieoesh15 than on node ieoesh09
Volume group /dev/vgqawt1jd1 is configured differently on node ieoesh09 than on node ieoesh15
Volume group /dev/vgqash20a is configured differently on node ieoesh28 than on node ieoesh15
Volume group /dev/vgqash20a is configured differently on node ieoesh15 than on node ieoesh28
Volume group /dev/vgqawt1dw1 is configured differently on node ieoesh28 than on node ieoesh15
Volume group /dev/vgqawt1dw1 is configured differently on node ieoesh28 than on node ieoesh09
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c30t1d0 on node ieoesh28 (25418018821308575088).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c32t1d0 on node ieoesh28 (25418018821308575088).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c30t1d3 on node ieoesh28 (25418018821308575314).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c32t1d3 on node ieoesh28 (25418018821308575314).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c30t12d1 on node ieoesh28 (25418018821299510160).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c32t12d1 on node ieoesh28 (25418018821299510160).
Volume group /dev/vgqawt1dw1 is configured differently on node ieoesh15 than on node ieoesh28
Volume group /dev/vgqawt1dw1 is configured differently on node ieoesh15 than on node ieoesh09
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c9t12d1 on node ieoesh15 (25418018821299510160).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c28t12d1 on node ieoesh15 (25418018821299510160).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c9t1d0 on node ieoesh15 (25418018821308575088).
Volume group /dev/vgqawt1dw1 on node ieoesh09 does not appear to have a physical volume corresponding to /dev/dsk/c28t1d3 on node ieoesh15 (25418018821308575314).
Volume group /dev/vgqawt1dw1 is configured differently on node ieoesh09 than on node ieoesh28
Volume group /dev/vgqawt1dw1 is configured differently on node ieoesh09 than on node ieoesh15
Volume group /dev/vgqawt1dw2 is configured differently on node ieoesh28 than on node ieoesh15
Volume group /dev/vgqawt1dw2 is configured differently on node ieoesh15 than on node ieoesh28
Volume group /dev/vgqawt1dw2 is configured differently on node ieoesh15 than on node ieoesh09
Volume group /dev/vgqawt1dw2 is configured differently on node ieoesh09 than on node ieoesh15
Volume group /dev/vgqawt1dw2a is configured differently on node ieoesh28 than on node ieoesh15
Volume group /dev/vgqawt1dw2a is configured differently on node ieoesh15 than on node ieoesh28
Volume group /dev/vgqawt1dw2a is configured differently on node ieoesh15 than on node ieoesh09
Volume group /dev/vgqawt1dw2a is configured differently on node ieoesh09 than on node ieoesh15

Do I need to clear all these message to get the config file?

Thanks

Sunil

[Matti_Kurkela]

The "Volume group ... is configured differently ..." messages might not be too serious, and I think they should not prevent the creation of the cluster configuration file.

But the "Volume group ... on node ... does not appear to have a physical volume corresponding to ..." messages are more serious: they might mean some LUNs are presented incorrectly. You should fix those, as they mean a package might be unable to move to the node that is missing some of package LUNs.

(Or if a certain package is not meant to ever go to a particular node, you should make sure that only the allowed nodes are explicitly listed in the package configuration file, instead of using "*" to allow all nodes. That should stop the warnings in that case.)

[Stephen Doud]

cmgetconf is more forgiving on newer releases (A.11.19) and will generate an outfile where as earlier versions stumbled on problematic issues such as VGs being configured differently across nodes.  Since no output file is being generated, you will have to address the issues.

For  either"configured differently" or "missing device" messages, the solution is usually the same - on the affected node, export the vg and reimport it, using a LVM map file that contains the VGID at the top, to allow LVM to discover ALL LUNs and all their PVlinks for a given VG.

1) Create an LVM map file:
$ vgexport -pvs -m <vgname>.map /dev/<vgname>

You can do the above command regardless of whether the VG is active or not.

The map file will contain 3 pieces of information:
Line 1: the unique VGID of the volume group.

Line 2+:   the lvol number      the lvol name (whether default or customized)

2) scp the map file to other nodes - and particularly to the node where it's /etc/lvmtab file does not accurately reflect the current list of PV/PVlinks to the LUNs in the VG.  Use /etc/lvmconf as the target directory

3) On the node where lvmtab is outdated:

ll /dev/*/group > /tmp/vggroup   # capture a list of major (LVM vs LVM2) and  minor numbers associated with each VG

vgexport <vgname>
mkdir /dev/<vgname>

mknod /dev/<vgname>/group c <major number> 0xNN0000  # where NN = minor number

vgexport -vs -m /etc/lvmconf/<vgname>.map /dev/<vgname>

4) Repeat for affected VGs.

5) For those systems that currently have the VG active, but not all pvlinks are listed in /etc/lvmtab

a. - identify all device files associated with the VG (including pvlinks)

b. vgextend /dev/<vgname>  /dev/*d*sk/<missing DSF->  ...

Where *d*sk can be:
cdisk:  cluster device special file

disk: agile address

dsk: legacy device special file

or, if using agile addresses

[Jagur]

Thanks for the updates..I would do reimport all the cluster VGs on error nodes and check..

Will update with results...