Product

• Brisque^® serve 4
• Timna

Platform

• 43P–150—Tiger2
• 44P–170—Hammerhead
• 6F1—Condor
• F50

Operating system

• IBM^® AIX^® 4.2.1, 4.3.3
• Legato 5.1, 6

Symptoms

A “WISS” error appears or a customer reports of such an error in HELIOS mail (during the daily check of Legato activity).

Solution

The solution to this problem involves the following sequence:

In Legato 5.1:

Do not begin with the Legato index recovery. Instead, execute nsrck and check if this command corrects the index. In Brisque 5 in particular, you should copy the indexes first into dataVolumes. Otherwise, if there is an error in nsrck, you might lose the entire online DB (Legato index), which will require mmrecov.
Note: nsrck requires less time and will save searching for the latest known good bootstrap.

Follow the steps below:

1. Create a folder legato in a dataVolume.

2. Back up your Legato database to a dataVolume. Type:
   # cd /backup/nsr
# /backup/bin/uasm -s -i ./index |
(cd /dataVolumes/ripro1.8.0/legato ;
/backup/bin/uasm -r -iY)
# /backup/bin/uasm -s -i ./mm |
( cd /dataVolumes/ripro1.8.0/legato ;
/backup/bin/uasm -r -iY )
# cp -p /backup/nsr/res/nsr.res
/dataVolumes/ripro1.8.0/legato
3. You can execute the following command while Legato processes are running or not.
   Note: You will be notified if they are not running.
   Type:
   # nsrck -F
   If the process is successful, the problem is solved.
   Only if the process fails, perform the following:
   Restore the original Legato directory from dataVolumes. Type:
   # cd /dataVolumes/ripro1.8.0/legato
# /backup/bin/uasm -s -i ./index | (cd /backup/nsr; uasm -r -iY)
# /backup/bin/uasm -s -i ./mm | ( cd /backup/nsr; uasm -r -iY )

Note: It is preferable to have a corrupt DB than not having any DB at all, which will then require you to restore a DB from an earlier date.

In Legato 6:

Check if WISS errors in daemon.log are accompanied by the following error:
daemon.002:01/19/04 16:30:14 nsrmmdbd: WISS error: Unable to mount /backup/nsr/mm/mmvolume6: bad database header
If yes, nsrck will not help, and you are advised to recover the indexes+media DB from the most recent reliable backup (prior to the date of the first corruption).

Technical contact(s)

Lev Weissman

Submitted by

Lev Weissman

Date submitted

09 Nov 2004