Cannot rebuild ir.probsummary records when their are only 1,3 or more nodes (instances) in a horizontally scaled envrionment

  • KM498414
  • 24-Sep-2008
  • 24-Sep-2008

Error

When doing a search on Incidents(probsummary) using the Description field (action/action) some words cause the client to crash. There are several words  that cause this crash, all seemingly random. This crash happens on both the web client and the fat client.
In sm.alert.log this line is logged when this happens.
RTE I Corruption-3-ir.probsummary, Corrupted IR index detected ; user(bob.Bob), application(cc.search), panel(select)

Cause

The error in the sm.alert.log indicates that IR Regen will help. However, when service manager configuration has 1, 3 or more instances the IR regen fails.

STR:

1. Setup HZ scaling with 1 or 3 nodes.

2. Go to the scirexpert file and purge all the data.

3. Go to any dbdict with an ir key and perform a ir regen.

4. Go review the scirexpert file. You will get a no data found.

5. Note that if you run the same tests with 2 nodes(instances)  the ir files are rebuilt correctly. This was tested with  4 nodes(instances) to see if it was an issue with even vs odd number of nodes. The ir files were not rebuilt with 4 nodes either.

Nothing is written to the log indicating an error even with irtrace parameters in the sm.ini file

Fix

SCR 42611  resolved inService Manager 7.02.