LMS 3.1 IPM collectors missing

[markschnabel]

I had set up several collectors and had them running, then checked IPM about a week later and nothing shows under collectors now. I logged into my source routers and the ipsla config is running. How can I get the configured collectors to show up in IPM again? or do I need to delete all the ipsla operations running and start over?

[Joe Clarke]

This may be related to a bug I recently found. Post your IPM Audit trail report.

[markschnabel]

According to the audit report, auto allocation was disabled on the 23. I dont believe we ever had it enabled. We added devices manually.I noticed all reports stopped on the 23 also.

[Joe Clarke]

Yes, this is my bug CSCtb00574. The bad news is, all of your collects and devices are gone from IPM. If you have a known good backup from before the problem occurred, you can restore that. Else, you will have to manually re-add your devices, then add back your collectors (or import them from the devices using the ipm CLI command).

As for fixing this so it doesn't happen in the future, a patch is available from TAC. What happened on the 23? Was there a power hit, or was the server rebooted abnormally?

[markschnabel]

Is this fixed in the 3.2 update? Thanks for the info.

I'm using a shadow router for a source and want to add remote site wan routers for responders. When a new site come on line, I want to add the device to the collector. It appears that I cannot edit the collector and need to delete and create a new new every time in order to add the new site.Is there a better way to do this?We have 5 ipsla operations to run against each site.thanks

[steve.sobot]

Hi Joe,

I actually have the same question as markschnabel - is the bug CSCtb00574 still present in the IPM released with CWLMS 3.2 ?

Since an upgrade to CWLMS 3.2 most of the collectors are no longer visible at one installation we are investigating.

Can you describe in a little more detail what the bugfix available through TAC actually does?  Thanks.

[Joe Clarke]

The bug is still present in IPM 4.2.  The fix prevents IPM from flushing the collector and device data if it detects that the DCR domain is null or empty.  This solves a corner case where IPM can find that the domain ID is null (e.g. on a power loss the data may not have been committed to the DB), and think that the DCR domain had changed.