stray IPMDB.db?

Unanswered Question
Nov 21st, 2008

Without fail, there's always a copy of IPMDB.db left in the / file system after restorebackup on LMS 3.1. Is this file needed?

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
yjdabear Fri, 11/21/2008 - 07:53

Is this an oversight in that it does this? We have limited space in /, so this runs the risk of filling it up.

Joe Clarke Fri, 11/21/2008 - 08:00

The old IPM data is cleaned up at the very end of the data migration process. However, there appears to be a bug with this. The code is looking for something like /backups/0/ipm/IPMDB.db. However, I believe the actual structure is /backups/0/ipm/database/IPMDB.db. However, I have no idea how a copy of IPMDB.db could end up in /.

yjdabear Fri, 11/21/2008 - 08:18

This sounds rather similar to my previous issue where was looking for ani.db in /backups/0/campus/database/, but had it in /backups/0/cmf/ani.db, because of messed-up /opt/CSCOpx/backup/manifest/cmf/database/cmf.txt and /opt/CSCOpx/backup/manifest/campus/database/ani.txt.

Here's the source LMS(2.6)'s /opt/CSCOpx/backup/manifest/ipm/database/ipm.txt

"ipm.txt" 2 lines, 44 characters



Joe Clarke Fri, 11/21/2008 - 08:25

This is right. A problem with the manifest file would also not explain how a copy winds up in /. But as I said, I believe there is a bug in which the old IPM database will not be cleaned up after migration.

Joe Clarke Fri, 11/21/2008 - 08:38

I would need to know the actual layout of the ipm backup directory first (i.e. an ls -alR).

Joe Clarke Fri, 11/21/2008 - 12:11

Yes, this is a bug, and I can provide a fix if you open a service request.

Joe Clarke Fri, 11/21/2008 - 12:22

No, and the more I look at the code, I don't think this fix is really needed. It certainly wouldn't clean up an IPMDB.db in /. Can you identify where exactly in the migration process this /IPMDB.db gets created?

yjdabear Fri, 11/21/2008 - 12:26

How precise does the identification have to be? Would the migration.log plus the timestamp of IPMDB.db suffice?

Joe Clarke Fri, 11/21/2008 - 12:29

Maybe. It would have to be fairly precise to track down the exact bit of code that is causing the problem.


This Discussion