OVERVIEW

 

A linter database verification script, TestLinterDB.pl, needs to be run when upgrading systems from 1.x -> 2.1.x -> 2.2.1 or higher.  The script was created because on some customer DXi systems corruption in the linter database was present and caused the upgrade to 2.2.1 to fail. 

 

To clean up the database and manually complete the upgrade after the fact required great effort and unnecessary pain, so the TestLinterDB.pl script was designed to be run prior to the upgrade, as part of the 2.2.1.x pre-upgrade procedure to determine if linter is intact and safe to upgrade.  See PTR 32928 for more detailed information.

 

If the script does not find any linter database corruption, it restarts services and you can proceed with the upgrade.  If the script detects corruption it should automatically repair the database and restart services and then you can proceed.  Either way, the output indicates if it is safe to proceed with the upgrade or not. 

 

The script outputs to the terminal window as well as to a log file located in /var/log/DXi/upgrades/testdb.<date>.log 

 

You can downloaded the script from csweb.

 

 

SYMPTOM / RESOLUTION

 

What to do if the script detects a problem and indicates you cannot proceed.  I’ve seen 2  different scenarios.

 

1) The script completes but provides a generic warning that you cannot upgrade until linter is repaired.  This is generally seen when trying to upgrade from 2.0.x.x to 2.2.1.2 or later.  This does not work because on the 2.0.x.x versions linter resides under /snfs but the script unmounts /snfs  first and can no longer access the database to run the check.  See PTR 34916

 

[root@NAS01 ~]# ./TestLinterDB.pl -t -r -s stopping heartbeat diagostic log file is: /var/log/DXi/upgrades/testdb.20130607_14:57:51.log

linter test indicates repair is required.

executing linter repair.

linter repair was not successful.

Do not upgrade until linter is repaired.

restarting heartbeat and exiting

Starting High-Availability services: [ok]

 

To get official workaround is to upgrade to 2.1.3 first and then to 2.2.1.2 or later.  A quicker method is to stop services, start cvfs and then run the script which will now be able to access the linter database.  It will indicate you can proceed with the upgrade unless there is a real problem with Linter.

 

# service heartbeat stop

#  service cvfs start

# service heartbeat status  

# TestLinterDB.pl  -t  -r  -s

 

2) If the script detects real errors like the ones below, linter needs to be repaired before upgrading.  Fixing the database involves several steps which will change depending on the specific errors seen.  These cases should escalated to SES for review.

 

Database Testing Report

------------------------

-  There are errors in 3 tables:

#31  *vdb*.*SL_EVENT_DETAILS*

#51  *vdb*.*SL_NOTIFICATION_EMAIL*

#115  *vdb*.*SL_NOTIFICATION_TRAP*

-  All errors can be fixed by SQL update file.

-  You should run TESTDB -s upd_file, and then execute the created  SQL update file through Linter Desktop or INL after starting Linter.

-  After that you should repeat the database testing procedure.

 

 

ADDITIONAL NOTES

The upgrade procedures call for running the TestLinterDB.pl script when upgrading to 2.2.1.x regardless of current version of fw.