cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Level 15

Orion 9.5 Database Maintenance

Jump to solution

Looks like the DB maintenance on my SLX server is crippling performance.  Since it started the last two days the CPU and disk IO are through the roof on the server.  This is causing false positives and outage alerts. 

I had a similar issue in 9.1 that there was a patch to fix which was referenced in this thread

Is anyone else having this issue?  I am going to take the same steps to disable the DB Maint until I have a workaround.

0 Kudos
1 Solution

There is a variable "@chunksize = 100" in the stored procedures that the DB maintenance uses this set a limit of 100 records per loop.  If you have a large amount of records to proces then this will take forever.

I ended up changing 3 of them to 100,000. (dbm_ResponseTime_DeleteStale dbm_InterfaceTraffic_DeleteStale dbm_InterfaceErrors_DeleteStale) After the change the first time the DB maintenance ran it took a few hour to complete but after that it now runs and completes in a reasonable amount of time

You may need to change more of the stored procedures depending of the size of your database but it seems any table over 400 MB will take a large amount of time to complete with the 100 row limit set.

View solution in original post

0 Kudos
32 Replies

SP1 for 9.5, where did you get it from?, I don't see it published yet.

0 Kudos

Please see this post

0 Kudos

the official response i got from them was to refer to this thread (great support!).  i set my chunk size to 1,000,000 in those three stored procedures and now everything runs smooth.

0 Kudos

Glad it worked and your welcome.

0 Kudos

That fixed the issue.  Changing the chucksize to 100000 on all three of the stored procedures.  The DBMaint only took about 20 minutes to complete which isn't bad since it hasn't ran since Sunday morning.

With it running daily I shouldn't run into this anymore. 

Hopefully Solarwinds will address this in the next SP or hotfix.  I will update more support case with this information.  Lasher thank you for your help

0 Kudos

Is this due to a change in 9.5?  never had this issue until we upgraded last week. 

0 Kudos
Level 9

There are a few procedures used in the Database Maintenance that are limited to 100 rows.  If it has a large number of rows to delete it has to loop over and over again.  This is what is slowing down the process.

Simple increases of that limit will speed up the process greatly.  I will have to look it back up all the ones that I found but one of them was "dbm_InterfaceErrors_DeleteStale".  (There were one or two others that have that same limit.)  After that limit was increased the Database Maintenance finished without issue for me.

Product Manager
Product Manager

Have you opened a support ticket to investigate?

0 Kudos

Not yet, I installed the upgrade Sunday afternoon so we had this issue early Monday AM and now this morning.  I just narrowed it down to DB Maint being the issue a few minutes ago.  I posted here first to see if anyone else had seen it in 9.5.  My next step is support.

The maintenance is hanging on summarizion data from Response Time.  My overnight employee tells me it had been on that section since 4 AM and I shut it down at 8 AM and it was still there.

0 Kudos

Opened case 98037 regarding this issue.  I will follow up here when I have a fix so others can apply if they have the same issue.

0 Kudos

I am having the same issue.  About an hour into database maintenance a process starts blocking and crushes the server.  I killed the process being blocked and the job is continuing to run.  Please post their solution when you have it.

0 Kudos


I am having the same issue.  About an hour into database maintenance a process starts blocking and crushes the server.  I killed the process being blocked and the job is continuing to run.  Please post their solution when you have it.



I have recieved some requests from support to gather some data which I will work on tomorrow.  I would hope you would also open a ticket so they can see your logs to help determine if it is something in the build.

0 Kudos