Implementing ServiceGuard extension for SAP

[Taurian]

Hi All,

I am deploying Serviceguard extension for SAP B.05.00. I want to make a two node cluster with one package. Serviceguard version installed is 11.18. HP-UX version is 11.23 (June 2008).

I am getting the following errors while trying to cmapplyconf:

# cmapplyconf -v -f -C /etc/cmcluster/PRD/clust1.conf -P /etc/cmcluster/PRD/dbciPRD.conf

Begin cluster verification...
Checking cluster file: /etc/cmcluster/PRD/clust1.conf
Note : a NODE_TIMEOUT value of 2000000 was found in line 160. This value
is recommended if the top priority is to reform the cluster as fast
as possible in case of failure. If the top priority is to minimize
reformations, consider using a higher setting. For more information see
the cluster configuration ASCII file or the Managing Serviceguard manual.
MAX_CONFIGURED_PACKAGES configured to 150.
Checking nodes ... Done
Checking existing configuration ... Done
MAX_CONFIGURED_PACKAGES configured to 150.
Gathering storage information
Found 25 devices on node ahlprddb
Found 25 devices on node ahlprdap
Analysis of 50 devices should take approximately 6 seconds
0%----10%----20%----30%----40%----50%----60%----70%----80%----90%----100%
Found 7 volume groups on node ahlprddb
Found 7 volume groups on node ahlprdap
Analysis of 14 volume groups should take approximately 1 seconds
0%----10%----20%----30%----40%----50%----60%----70%----80%----90%----100%
Detected node ahlprdap on revision A.11.18.00 that does not support newer network discovery. Will revert to older network discovery.
Gathering network information
Beginning network probing (this may take a while)
Detected node ahlprdap on revision A.11.18.00 that does not support newer network probing (including cross-subnet probing). Will revert to older network probing.
Completed network probing
Parsing package file: /etc/cmcluster/PRD/dbciPRD.conf.
Attempting to add package dbciPRD.
Failed to validate /etc/cmcluster/scripts/mscripts/master_control_script.sh.

On node ahlprddb validation failed with:
[0] /etc/cmcluster/scripts/sg/volume_group.sh
[1] /etc/cmcluster/scripts/sg/filesystem.sh
[2] /etc/cmcluster/scripts/sg/package_ip.sh
[3] /etc/cmcluster/scripts/sgesap/dbinstance.sh
[4] /etc/cmcluster/scripts/sgesap/sapinstance.sh
[5] /etc/cmcluster/scripts/sg/resource.sh
Entering SGeSAP dbinstance.sh 'validate' runtime steps ...
SGeSAP - B.05.00.00 - Last modified 02/11/08
Found: /etc/cmcluster/scripts/sgesap/sap_functions.sh
Found: /etc/cmcluster/scripts/sgesap/customer_functions.sh
Not found: /etc/cmcluster/scripts/ext/dbciPRD_customer_functions.sh
(check_versions): ERROR: Mismatched Serviceguard version "A.11.18.00" (via /usr/sbin/cmversion).
(check_versions): ERROR: You need at least A.11.19.
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/dbinstance.sh
Entering SGeSAP sapinstance.sh 'validate' runtime steps ...
SGeSAP - B.05.00.00 - Last modified 02/11/08
Found: /etc/cmcluster/scripts/sgesap/sap_functions.sh
Found: /etc/cmcluster/scripts/sgesap/customer_functions.sh
Not found: /etc/cmcluster/scripts/ext/dbciPRD_customer_functions.sh
(check_versions): ERROR: Mismatched Serviceguard version "A.11.18.00" (via /usr/sbin/cmversion).
(check_versions): ERROR: You need at least A.11.19.
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/sapinstance.sh


cmapplyconf: Error found in package file: /etc/cmcluster/PRD/dbciPRD.conf.

[Rita C Workman · Accepted solution]

NODE_TIMEOUT value of 2000000 was found in line 160. This value
is recommended if the top priority is to reform the cluster as fast
as possible in case of failure. If the top priority is to minimize
reformations, consider using a higher setting.
>>> Try changing the value to 6000000

ERROR: Mismatched Serviceguard version "A.11.18.00" (via /usr/sbin/cmversion).
(check_versions): ERROR: You need at least A.11.19.
>>>It appears you have conflicting versions installed. Get everything on the same version level.

Found: /etc/cmcluster/scripts/sgesap/customer_functions.sh
Not found: /etc/cmcluster/scripts/ext/dbciPRD_customer_functions.sh
>>>You have multiple files that it can't find. Make sure the run command scripts are where you stated.

Something to get you started,
Rita

[Taurian]

Hi Rita,

Thanks for the prompt reply.

Fixed the timeout and dbciPRD_customer_functions.sh errors.

Regarding rest of the issues:

1. Regarding compatibility info, the release notes for SGeSAP lists compatibility with Serviceguard 11.18. Then why am I getting this error. Also, how do I upgrade to Serviceguard version 11.19?

2. Regarding the errors:
Failed to validate /etc/cmcluster/scripts/mscripts/master_control_script.sh
and
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/sapinstance.sh
I have verified that these scripts are in the place mentioned in the above path. Then why the errors? The file permission are also set to readable and executable by everyone.

[Doug O'Leary]

Hey;

The way I'm reading this: the SAP extensions that you installed is looking for MCSG version 11.19 while you have ver 11.18.

You should be able to do a verbose listing on the sap extensions to find out if that's accurate.

Doug

[Taurian]

Hi Doug,

Thanks for the reply.

Can you please share the command?

[Taurian]

Hi all,

Taking it one step at a time....I decided to just build a cluster first, and build the package later on.

Issuing command: "cmapplyconf -v -C /etc/cmcluster/PRD/clust1.conf" on node1 has successfully build a cluster.

1. Do I need to run cmapplyconf command on node 2 as well? As the output of cmviewcl shows node2 status unknown.

2. The cmruncl command is not completing. node2 status is unknown and node1 status is reforming. After a while the cluster and both nodes show the status down.

3. cmviewcl command on node2 gives error:
Unable to open communications to configuration daemon: Permission denied
Unable to connect to configuration database.
Also, the cmcld deamon is not running on node2.

4. Syslog of node2 shows the following error:
Jan 12 21:11:10 ahlprdap cmclconfd[4698]: Unable to properly gather the remote user for fd 3. Please make sure the remote node is running identd.
Request from root on node ahlprdap to start the cluster on this node failed: not authorized.

[Taurian]

Hi All,

I've successfully formed a cluster, but I am having trouble with applying package configuration. I am getting the following error:

#cmapplyconf -P /etc/cmcluster/PRD/dbciPRD.conf
Failed to validate /etc/cmcluster/scripts/mscripts/master_control_script.sh.

On node ahlprdap validation failed with:
[0] /etc/cmcluster/scripts/sg/volume_group.sh
[1] /etc/cmcluster/scripts/sg/filesystem.sh
[2] /etc/cmcluster/scripts/sg/package_ip.sh
[3] /etc/cmcluster/scripts/sgesap/dbinstance.sh
[4] /etc/cmcluster/scripts/sgesap/sapinstance.sh
[5] /etc/cmcluster/scripts/sg/resource.sh
/etc/cmcluster/scripts/mscripts/master_control_script.sh[121]: /etc/cmcluster/scripts/sgesap/dbinstance.sh: Execute permission denied.
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/dbinstance.sh
/etc/cmcluster/scripts/mscripts/master_control_script.sh[121]: /etc/cmcluster/scripts/sgesap/sapinstance.sh: Execute permission denied.
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/sapinstance.sh

On node ahlprddb validation failed with:
[0] /etc/cmcluster/scripts/sg/volume_group.sh
[1] /etc/cmcluster/scripts/sg/filesystem.sh
[2] /etc/cmcluster/scripts/sg/package_ip.sh
[3] /etc/cmcluster/scripts/sgesap/dbinstance.sh
[4] /etc/cmcluster/scripts/sgesap/sapinstance.sh
[5] /etc/cmcluster/scripts/sg/resource.sh
Entering SGeSAP dbinstance.sh 'validate' runtime steps ...
SGeSAP - B.05.00.00 - Last modified 02/11/08
Found: /etc/cmcluster/scripts/sgesap/sap_functions.sh
Found: /etc/cmcluster/scripts/sgesap/customer_functions.sh
Found: /etc/cmcluster/scripts/ext/dbciPRD_customer_functions.sh
(check_versions): ERROR: Mismatched Serviceguard version "A.11.18.00" (via /usr/sbin/cmversion).
(check_versions): ERROR: You need at least A.11.19.
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/dbinstance.sh
Entering SGeSAP sapinstance.sh 'validate' runtime steps ...
SGeSAP - B.05.00.00 - Last modified 02/11/08
Found: /etc/cmcluster/scripts/sgesap/sap_functions.sh
Found: /etc/cmcluster/scripts/sgesap/customer_functions.sh
Found: /etc/cmcluster/scripts/ext/dbciPRD_customer_functions.sh
(check_versions): ERROR: Mismatched Serviceguard version "A.11.18.00" (via /usr/sbin/cmversion).
(check_versions): ERROR: You need at least A.11.19.
ERROR: Failed to validate /etc/cmcluster/scripts/sgesap/sapinstance.sh


cmapplyconf: Error found in package file: /etc/cmcluster/PRD/dbciPRD.conf.



Please guide as soon as possible!

[Rita C Workman]

First, take another look at Doug's comments on the SAP vs SG versions.

On the errors, it's the spaghetti kind of errors, but it is telling you exactly what you need to fix:

Package dbciPrd.conf
Failed to validate /etc/cmcluster/scripts/mscripts/master_control_script.sh.
>>>is that file really there and are the permissions set right. Check cause sometimes it's just a typo in the conf file on the filename that is so easy to not see.

On node ahlprdap it is complaining again about permissions on files--check the permissions and check that file exists.

On node ahlprddb
>>>this is what Doug was referring to on software versions loaded that needs to get resolved.

Service Guard is somewhat demanding on wanting everything 'just so'.

Rgrds,
Rita

[Taurian]

Hi Rita and Doug,

Your comments have been great help. I have done following for the two issues I was facing.

For the Permission Denied issue, I assigned "chmod 755" permissions to the SG scripts on node2. That resolved that issue.

For the compatibility issue (which I still don't understand as both SGeSAP B.05.00 and Serviceguard 11.18 are compatible to each other) I found the script in which it was comparing the version via cmversion command. Turns out that it is the sap_functions.sh script. I changed the value of the variable MIN_SGVERSION to 11.18, and voila....everything started working!

I have now built a package which is starting and mounting the filesystems correctly.

I'll keep you posted on errors as I proceed.

[Felix Eng]

Check your version conflictâ s, may be version error due to your toolkit release for further also check start, stop instance scripts.

is there any NFS involvement?