none
Troubleshooting high CPU usage store.exe and w3wp.exe

    Question

  • We have a fully patched Exchange 2010 SP2 system in a single server setup, We are seeing very high CPU usage on store.exe (30-60 percent at times) and w3wp.exe (20% or so) These sometimes cause the CPU usage to spike to 100 % causing performance degredation.
    Our event logs appear to be clean of serious looking errors. The only ones I found are pasted below. The MAPI limit errors tend to reoccur.
    I've already run the online database maintenance tools on both mailbox databases with no fixes or issues reported.

    Log Name:      Application
    Source:        MSExchangeIS
    Date:          2/1/2012 3:30:46 PM
    Event ID:      9646
    Task Category: General
    Level:         Error
    Keywords:      Classic
    User:          N/A
    Computer:      exchsrvr
    Description:
    Mapi session "fc9d36b3-c892-4b06-b82b-35a1406c6bc0: /o=orgname/ou=First Administrative Group/cn=Recipients/cn=lengle.3" exceeded the maximum of 500 objects of type "objtFolder".


    Log Name:      Application
    Source:        MSExchangeIS
    Date:          2/2/2012 12:01:40 PM
    Event ID:      9646
    Task Category: General
    Level:         Error
    Keywords:      Classic
    User:          N/A
    Computer:      exchsrvr
    Description:
    Mapi session "54d687ba-70c3-4783-bb75-c6bfa427ed8b: /o=orgname/ou=First Administrative Group/cn=Recipients/cn=ray.30" exceeded the maximum of 250 objects of type "objtMessage".


    Log Name:      Application
    Source:        MSExchange ADAccess
    Date:          2/2/2012 10:35:41 AM
    Event ID:      2915
    Task Category: General
    Level:         Error
    Keywords:      Classic
    User:          N/A
    Computer:      exchsrvr
    Description:
    Process w3wp.exe () (PID=4060). User 'Sid~RF\ray.30~EWS~false' has gone over budget '463' times for component 'EWS' within a one minute period. Info: 'Policy:DefaultThrottlingPolicy_9e9f3ff5-863c-426f-9c97-762da29f0b07, Parts:CAS:462;'. Threshold value: '100'.

    Where should we go from here to troubleshoot this issue?

    Friday, February 03, 2012 3:32 PM

All replies

  • What are the specs for the serer and are they adequate for your user profile?

     

    Friday, February 03, 2012 3:41 PM
    Moderator
  • The server is virtualised on an VMWare infrastructure. It has 16 GB of RAM, plenty of disk space ~2TB and has 2 dedicated CPU cores zeon 2.9GHz.

    When the CPU spikes, the store.exe also consumes a large amount of RAM !8GB. This doesn't happen all the time but is relatively constant.

    There are only ~300 users on this server and historically our performance numbers have been about a third of what they are now.

    Friday, February 03, 2012 3:51 PM
  • Can you bump up to 4CPU?

    Even with 300 users, 2 CPU never seems to cut it for a MBX/CAS.

     

    Friday, February 03, 2012 3:53 PM
    Moderator
  • We'll investigate increasing the CPU allocation, however this has worked for us in the past. it is only very recently that this issue has occured. We also process inbound and outbound email through ironport spam appliances so the load on the HT part should be limited to actual inbound and outbound mail with very little spam.
    Saturday, February 04, 2012 12:16 AM
  • We'll investigate increasing the CPU allocation, however this has worked for us in the past. it is only very recently that this issue has occured. We also process inbound and outbound email through ironport spam appliances so the load on the HT part should be limited to actual inbound and outbound mail with very little spam.

    I dont think its the handling of mail that cause performance issues, more likely the work the CAS/MBX is doing handling all the client conenctions store activity.

    When did this work before in the past? Can you clarify that?

    Saturday, February 04, 2012 1:28 PM
    Moderator
  • We only noticed the increased CPU usage after the January patch window. However, the server ran fine for 4 or 5 days after the patches were applied before this issue popped up. We did also have an event where the storage that holds our Exchange Databases was briefly disconnected. However after rebooting the server with the storage reconnected this issue was resolved. And the high CPU started before the storage event.
    Sunday, February 05, 2012 4:01 PM
  • Hi,

    What is the issue now?

    event 9646 and event 2915 can be found in the event log?

    Please try to use task manager to verify the cpu that store.exe eats. Also  you can try to use Process ExPlorer to verify the issue.


    Xiu Zhang

    TechNet Community Support

    Tuesday, February 07, 2012 8:18 AM
    Moderator
  • Those are the only errors in our logs. Store.exe and W3wp.exe continue to periodically spike causing CPU usage to be 100%.

    Tuesday, February 07, 2012 12:11 PM
  • Hi,

    You can find some instructions here, can help you to solve this issue.

    http://social.technet.microsoft.com/Forums/fi-FI/exchange2010/thread/4731aae5-b367-4cc9-a310-04e3defa98aa

    Regards,

    Tuesday, February 07, 2012 1:32 PM