Server Management - Systems Insight Manager
1748239 Members
3851 Online
108759 Solutions
New Discussion

Re: mxdomainmgr memory usage high

 
Kerry Quillen
Regular Advisor

mxdomainmgr memory usage high

I just upgraded my SIM 6.3 to 7.0 and performance has gone down the tubes.  Mxdomainmgr is using 95% of memory.  I can stop the process, it restarts, and within an hour it will be back to 95%.  My server is a vm running Win2008 R2 x64 with 10gb ram.  My sql db is local and I also have VCRM installed.  I am managing approx 700 devices.  I did change the account password that SIM uses prior to the upgrade and made all of the necessary service account, DB, etc changes and everything seemed to be fine.

 

Prior to the upgrade, mxdomainmgr memory usage would creep up over time but not so fast - usually took about a month.  I've tried numerous suggestions that I have found online w/ no luck.  Just downloaded the performace tuning doc and will give that a read this weekend.    

 

Nothing that means anything to me in the logs or event viewer.  Unless I can find a solution, I may reinstall SIM and put db on seperate server as well as, VCRM. 

 

Any other suggestions?

 

 

5 REPLIES 5
jim goodman
Trusted Contributor

Re: mxdomainmgr memory usage high

I have been hearing of a memory hit with 7.0 - there is a lot more going on under the hood that a bump in memory should cure.

 

700 in a VM with a local SQL is going to be taxing as it is, you may want to move SQL Server off that VM. I am assuming you are using full SQL Server and a 64-bit VM.

shocko
Honored Contributor

Re: mxdomainmgr memory usage high

I would check the following as general guideline in any version of SIM where memory usage is ramping up:

 

  1. Numberofmanagednodes
  2. Pollingfrequencyoftasks. Tryandminimizetherunningofunneededtasksandminimizetheamountofthem
  3. Makesureyouaredeletingeventolderthansay90days. TheDBwillswellifyoudon'tclearthemdown. Ionlykeep30daysworthonline
  4. Checkyourdiscoverytasksandmakesuretheyarenotrunningtoofrequently
  5. Check the VCRM and make sure its working. I have moved this off to another server in the past to ease load on the disks.
  6. Thick provision your .vmdks
  7. Remove an uneeded hardware form the VMs
  8. Check the memory ballooing of the hypervisor
  9. Check CPU time splicing if using multiple CPUs on the VM

 

If my post was helpful please award me Kudos! or Points :)
Kerry Quillen
Regular Advisor

Re: mxdomainmgr memory usage high

I am using full SQL and 64bit vm.  I recall seeing a doc that explained how to move the db to its own server but cannot locate it now.  Any idea where I can find that?  Thx

jim goodman
Trusted Contributor

Re: mxdomainmgr memory usage high

Here is the doc for moving the SQL DB

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c03225518/c03225518.pdf

 

Shocko brings up some good points to consider and may be worth investigating. Typically in a vm once I get to where I need full SQL Sever I kick the db to another host as I do also the VCRM and Repository. If you use Insight Remote Support that adds PostgreSQL locally which you can't move and it can really chew up performance put it all together and eventually mxdomainmgr can eventually make your CMS unusable.

 

Also make sure your uncleared events are routinely cleared and that you don't have a ton of tasks stepping on themselves like mammoth event handlers, auto-discoveries, and your regularly scheduled tasks. All can impact the other by a continual daisy chain of never ending tasks or even worse individual tasks that never finish.

 

It would be great if you report back how it went and what worked for you. Keep in mind your mileage may vary :)

shocko
Honored Contributor

Re: mxdomainmgr memory usage high

Sorry guys, the formatting of my last post got messed up so here it is again in readable format!

 

I would check the following as general guideline in any version of SIM where memory usage is ramping up:

 

  1. Number of managed nodes
  2. Polling frequency of tasks. Try and minimize the running of unneeded tasks and minimize​the amount of them
  3. Make sure you are deleting event older than say 90 days. The DB will swell if you don't clear them down. I only keep 30 days worth online
  4. Check your discovery tasks and make sure they are not running​g too frequently
  5. Check the VCRM and make sure its working. I have moved this off to another server in the past to ease load on the disks.
  6. Thick provision your .vmdks
  7. Remove an unneeded hardware form the VMs
  8. Check the memory ballooning of the hypervisor
  9. Check CPU time splicing if using multiple CPUs on the VM
If my post was helpful please award me Kudos! or Points :)