NPM v8 Database maintenance log questions.

Afternoon,

Our NPM database's nightly maintenance is taking an age (nearly 24 hours!) to complete (Nodes 549, Interfaces 1700 and Volumes 0 - yes I know that's quite a lot).  I've looked through the log and there are two occasions where a particular step takes ages to complete; one is 1.5 hours and the other is 5.5 hours:

01/07/2007 18:03:09        Deleting all Response Time statistics without a parent Node
01/07/2007 19:38:02        Step Completed Sucessfully

01/07/2007 19:55:59        Deleting all Interface Traffic statistics without a parent Interface
02/07/2007 01:24:13        Step Completed Sucessfully

 
I don't understand what the 'without parent Node/Interface' bit means.  How do I find out what the problem is, assuming there actually is one?

Regards,

 Kris.
 

The 'without parent Node/Interface' bit is means Orion is deleting stats for interfaces that the "parent" node has been removed.  Have you removed large switches recently?  From your post it looks as if you are not on Orion v8.1, so I would recommend you open the database manager and perform a manual compact/re-build index on your largest tables, if not all of them.  This will take a long time to do, but it is very much worth the effort.  Now this is only a suggestion

Thanks for the reply John.  We are using Network Performance Monitor from the Toolset version 8, not Orion.  I've done the Database 'Compact and Repair' a couple of times and it brings the database size down a lot but the 'Nightly Maintenance' still takes an age.  Is the same/similar thing you suggest to try?

We've not removed any switches either.  Surely if you delete a switch NPM should then delete any child interfaces?

 
Regards,

Kris. 

I believe the inner workings of the toolset NPM is similar to the Orion NPM, so if you have already preformed the manual Compact/Rebuild Indexes and you still have the lengthy Nightly maint issue, have you checked the Anti virus software running on the machine and make certain that the db itself is not being scanned on access?  When was the last time you shutdown the NPM and Defragmented the drive that the db is on?

The data is held in 2 tables

Table 1 has node information (id, nodename, syslocation etc)

Table 2 has response time info (response time etc) and a link to the node table so that all response times are linked to the correct node

If a node is deleted from Table 1 then all the records in Table 2 are left with a link to nothing ie they have no parent

 Are you deleting nodes or interfaces on a regular basis?

Thanks for the replies John and Darryld.

John: I don't think we have shutdown NPM AND defrag'd the drive in ages.  I've just done that now and there were about 1000 fragments, so quite a few.  NPM does get shutdown weekly and a db 'compact and repair' is run to bring the file size down - we hit the 2GB limit quite often otherwise.  We'll see how that changes things over the weekend.

Darryld:  We don't delete nodes or int.'s very often.  Is there an easy way to make sure all entries in Table2 have a link to a valid node in Table1?  Or do I have to manually check using Access?

Regards,

Kris.
 

Kris,

Which format is your DB in (MS Access or Netperfmon native) and could you have another machine with an ODBC connection into it while the maint job is running?  Just throwing ideas out here...

Hi John,

The ideas are welcome :-)

I didn't know there were different DB formats that you can use.  How do I check?  The file name is SWNetPerfMon.cfg and I can open it in MS Access...  No other computers connect to our NMS so there shouldn't be any remote [ODBC] connections.

We use the NPM scheduler for the nightly maintenance so I guess NPM is still running while it's doing this.  Is that [inbuilt function] likely to cause a problem?

Regards

Kris.
 

oops, I was thinking of a Network Sonar DB...  I've had a cup or two of coffee now :)

If your using the Nightly maint function under settings there "should be" no conflict.