how to start mcsg with rac package ?cmrunpkg ?

[Sammy_2]

How do you run the darn rac package ?

All the package does
is activate vg (ops) in shared mode and starts oracle I can do all this manually (see below) but "cmrunpkg ojrdhes5"
command yields all kinds of messages. The other node the vg is activated in server mode.
I have tried activating the vg in shared mode,also deactivating th vg, commenting out "start oracle" line, but cant get
cmrunpkg to run successfully. (SEE BELOW)

excerpts from control file
===================
VGCHANGE="vgchange -a s"
...
..
.

: # do nothing instruction, because a function must contain some command.
# START ORACLE
#/app/oracle/admin/scripts/dbboot start all

==========================



#cmrunpkg inbphes5sg
cmrunpkg : Node inbphes5 is currently unable to run package inbphes5sg.
Check the syslog on node inbphes5 and pkg log files for more detailed information.

# vgchange -a n ops
Deactivated volume group in Shared Mode.
Volume group "ops" has been successfully changed.

# cmrunpkg inbphes5sg
cmrunpkg : Node inbphes5 is currently unable to run package inbphes5sg.
Check the syslog on node inbphes5 and pkg log files for more detailed information.

# vgchange -a s ops
Activated volume group in Shared Mode.
This node is a Client.
Volume group "ops" has been successfully changed.

# cmrunpkg inbphes5sg
cmrunpkg : Node inbphes5 is currently unable to run package inbphes5sg.
Check the syslog on node inbphes5 and pkg log files for more detailed information.


# tail syslog.log

rt the package
Jul 29 16:29:49 inbphes5 cmcld: Unable to start package inbphes5sg. Node inbphes5 not enabled to run it.
Jul 29 16:29:56 inbphes5 LVM[5684]: vgchange -a n ops
Jul 29 16:30:00 inbphes5 CM-CMD[5685]: cmrunpkg inbphes5sg
Jul 29 16:30:00 inbphes5 CM-CMD[5685]: Request from root on node inbphes5 to start the package
Jul 29 16:30:31 inbphes5 CM-CMD[5698]: cmrunpkg inbphes5sg
Jul 29 16:30:31 inbphes5 CM-CMD[5698]: Request from root on node inbphes5 to start the package
Jul 29 16:31:01 inbphes5 LVM[5703]: vgchange -a s ops
Jul 29 16:31:05 inbphes5 CM-CMD[5706]: cmrunpkg inbphes5sg
Jul 29 16:31:05 inbphes5 CM-CMD[5706]: Request from root on node inbphes5 to start the package
inbphes5:/var/adm/syslog ==>










privacy statement using this site means you accept its terms


© 1994-2005 Hewlett-Packard Company

[Mohanasundaram_1]

Hi Sammy,

I understand your frustration. I went through the same when I first configured this :-)

The problem is most likely at the "do nothing" line.

But provide us the package ascii file and the control file here. I will recollect the exact problem then.

With regards,
Mohan.

[Sammy_2]

Attaches is control file. ascii file is the next response.
I tried various command again but to no avail. #vgchange -a s ops works but I want
cmrunpkg to do that so cmviewcl shows that the pkg is running.

# vgchange -a n ops
Deactivated volume group in Shared Mode.
Volume group "ops" has been successfully changed.

# vgchange -c y ops
vgchange: Failed to set the specified attributes for volume group "/dev/ops".

# cmrunpkg inbphes5sg
cmrunpkg : Node inbphes5 is currently unable to run package inbphes5sg.
Check the syslog on node inbphes5 and pkg log files for more detailed information.

# vgchange -a s ops
Activated volume group in Shared Mode.
This node is a Client.
Volume group "ops" has been successfully changed.

# cmrunpkg inbphes5sg
cmrunpkg : Node inbphes5 is currently unable to run package inbphes5sg.
Check the syslog on node inbphes5 and pkg log files for more detailed information.

[Sammy_2]

Attaches in package config. file

Thanks. see if you can figure out why package will not start

[Thayanidhi]

Hi,
What does the controllog file shows? (located in the same folder as control script)
I think the issue may be, once you set vg as shared and cluster aware (-S y -c y) then don't activate your self. Deactivate (-a n) and start the package. The package control script would activate the VG.

Did you try this?

control.sh log file would give more details why the pkg start fails.

Regds
TT

[Mohanasundaram_1]

Hi Sammy,

In my case the problem was in the conf file

RUN_SCRIPT /etc/cmcluster/inbphes5sg/inbphes5sg.sh
RUN_SCRIPT_TIMEOUT NO_TIMEOUT
HALT_SCRIPT /etc/cmcluster/inbphes5sg/inbphes5sg.sh
HALT_SCRIPT_TIMEOUT NO_TIMEOUT


I had defined the run script file but the script itself had only a line

:

Which was the problem. Check what is the content of this script at /etc/cmcluster/inbphes5sg/inbphes5sg.sh

My problem was, the VG will get activated manually but when the package is started it does not get activated.

With regards,
Mohan.

[melvyn burnard]

One thing to check is the package log. What reason does that give for not running the package?
Also, use cmviewcl -v -p package_name to check the current switching parameters for the package.

[Matti_Kurkela]

A message like:
"cmrunpkg : Node inbphes5 is currently unable to run package inbphes5sg"
usually means that there has been a problem with the package. Either the run script, the halt script or a service in the package has failed.

Any of these conditions can cause ServiceGuard to mark the package as "don't try to run on this node, it didn't work last time".

See this table:
http://docs.hp.com/en/B3936-90079/ch03s04.html#babcbaeh

You can check the state of the package with "cmviewcl -v -pinbphes5sg". You'll see "disabled" in the SWITCHING column for node inbphes5.

When the problem that originally caused the package to fail is fixed, you must re-enable the package to run on the node that had the problem.

In your case, the command would be
cmmodpkg -e -n inbphes5 inbphes5sg

[Sammy_2]

I just rebooted and the pkg started ok. Not the best solution but I will keep an eye on the pkg to see
if it happens again/

This is 11.16 so the command below does not work.
# cmviewcl -c -p inbphes5sg
cmviewcl: Bad option inbphes5sg
usage: cmviewcl [-v] [-r {A.11.09|A.11.12}] [-c cluster_name] [-n node_name]... [-l {package|cluster|node|group}]

Matti may have a very close to what i wanted. some good links too.
Thavandi has a good point too abt deactivation vg b4 activating in shared mode.

Thanks to all.

[Sheriff Andy]

Sammy,

You can use the cmviewcl w/ no options & that will give you the cluster status. Otherwise you can use cmviewcl -v.

Next time this happens, you might try manually bringing up the disks.

vgchange -a e vg_group
mount disks
verify they are operational
umount
vgchange -a n vg_group

I was also wondering if the node that you were trying to mount to was active. If when you do the cmviewcl & it shows the node is down. You can do a cmrunnode -n inbphes5.

andy