M/C Service Guard A.11.11.16 - Pkg startup problem( ORACLE_TRACE)

[S.M.SANKAR KUMAR NADAR_2]

We hav 3*Nos of HP rp34xx Servers in single cluster.

We are using Oracle 9i for Database

Current problem is with the "pkg" as it is not starting . It terminateds with the following error " dbstart[18]: ORACLE_TRACE: Parameter not set.
logout
ERROR: Function customer_defined_run_cmds
ERROR: Failed to RUN customer commands"

due to the above error the below error also generated

dbshut[18]: ORACLE_TRACE: Parameter not set.

OS : HP UX 11i /64 Bit

Since this is the 1st time we are getting the error what could be the cause

The error was found in thue file "/etc/cmcluster/pkg_db/pkg_db.sdf.sh.log"

Log file attached FYR

[Steven E. Protter]

Log says this:

/dev/vx/dbdg/user01
/dev/vx/dbdg/user03
vxfs fsck: /dev/vx/dbdg/ruser01:cannot stat /dev/vx/dbdg/ruser01
ERROR: Function check_and_mount
ERROR: Failed to fsck one of the logical volumes.
vxfs fsck: /dev/vx/dbdg/ruser03:cannot stat /dev/vx/dbdg/ruser03
ERROR: Function check_and_mount
ERROR: Failed to fsck one of the logical volumes.


Looks like a logical volume can't be mounted.

That could mean a few things:

1) The logical volume's volume group is activated on another cluster.
2) The logical volume has a problem and needs fsck run on it.

Please correct the logical volume issue and try again. It may work at that point or generate a new error message.

SEP

[Warren_9]

any error when u run the dbshut/dbstart manually??

[S.M.SANKAR KUMAR NADAR_2]

If i run the dbstart/dbshut manually also I get the same error.

At the same time If I mount the volumes of "dbdg" group manually - there is no error & I can start the Oracle Database Manually without any issues.

But When I run the "cmruncl" it is exiting from

dbstart[18]: ORACLE_TRACE: Parameter not set.
logout
ERROR: Function customer_defined_run_cmds
ERROR: Failed to RUN customer commands....

[Ermin Borovac]

It seems like dbstart does su to oracle (su - ) account in order to start the database.

This will source .profile of oracle user and default profile has the following line.

set -u

It causes unset variables to be treated as error. ORACLE_TRACE, which is referenced by oracle startup script, is not set thus the error.

[S.M.SANKAR KUMAR NADAR_2]

Hi Thanks for Everyone

Currently the problem was resolved . The details as below.

The Problem was with the following commands in the Control (ASCII) file of the cluster(/etc/cmcluster/pkg_db/pkg_db.sdf.sh)

In the customer defined Functions part - the commands (Which was creating problem)

"su - oracle9i dbstart"
"su - oracle9i dbshut"

Which is now changed as below

"su - oracle9i -c dbstart"
"su - oracle9i -c dbshut"

& tested the autostartup of Oracle 9i- Working fine

Cluster also running with the pkg & working fine

[renarios]

Hi,

Be aware that the dbstart an dbshut loop throug the /etc/oratab file and try to startup/shutdown every database which is made "active" (if third field is 'Y'). So, having a failover cluster with all oratab files synchronized, the dbstart/shut commands will give plenty errors if the databases are spread over the cluster.
My advice is to create a "cluster aware" script to start and stop all databases which packages are active on that server.

Cheerio,

Renarios