cluster pkg problem

[Aungshuman Paul]

Hi,

Can anyone help me regarding the following issues :

# cmrunpkg -v -n erpdb1 sapdbci
Running package sapdbci on node erpdb1
The package script for sapdbci failed with no restart. sapdbci should not be res tarted
Unable to run package sapdbci on node erpdb1
Check the syslog and pkg log files for more detailed information
cmrunpkg: Unable to start some package or package instances

N.B. For more details I attacg the syslog and package log file

[likid0]

On you package log:

Mar 27 01:19:05 AM - Node "erpdb1": Deactivating volume group vgerpprdraid1
vgchange: Couldn't deactivate volume group "vgerpprdraid1":
Device busy
Mar 27 01:19:08 AM - vgchange -a n vgerpprdraid5 failed, trying again.
vgchange: Couldn't deactivate volume group "vgerpprdraid5":

Looks like you have mount points active from vgerpprdraid5 and vgerpprdraid1.

you should umount the fs that are mounted that belong to those vgs, and then vgchange -a n vgs, and then try to start the package


also you have this error all the time, the cluster custom run script can't find the files:

/usr/lbin/pkg_sapdbci[2]: /usr/sap/PRD/SYS/exe/run/startsap.sh: not found

[Duncan Edmonstone]

a couple of things here:

i) this appears to be a SAP cluster, but at first glance it looks like you aren't using Serviceguard Exetnsion for SAP... any reason for that?

ii) Looks like some of the volume groups are not correctly configured for cluster use:

- Make sure all the filesystems associated with the cluster package are un-mounted.
- Make sure all the volume groups associated with the package are de-activated (vgchange -a n vgname)
- Make sure all VGs associated with the cluster package are defined as cluster VGs (vgchange -c y vgname)

Now try and restart the package...

HTH

Duncan

[Sajjad Sahir]

Dear Aung

u have to umount the file system first
then do volume group deactivation,

vgcahnge -a n /dev/vg##

then try to start in volume group in cluster mode

vgchange -c y /dev/vg## ok

if it is in sharing u have to use

vgchange -S y -c y /dev/vg##

if it is not ok please send me cmviewcl -v output.

thanks and regards

Sajjad Sahir

[UVK_1]

I am not able to download the attachment, not sure why :(

Did u ensure the cluster controlled VG's or mounts are not configured in /etc/fstab, ensure none of those mounts used by your packed are mounted while you start the package.

When ur package is down, Try to manually activate and deactivate the VG you had problems with as per the log and see if are able to do it. You might want to fix the problem you face by doing so.

Did u do any changes with the package configuration recently. If yes..check your config to see if you have missed any vg entries

My comments are without seeing the logs. Nevermind them if they are of no use .

[Aungshuman Paul]

Thanks, My problem is solved by engaging HP's backline engineer. As I know It was a problem in cluster control packages.

Aungshu