Forum Discussion

WiTSend's avatar
WiTSend
Level 6
11 years ago

EV11 w3wp memory issues

We've seen memory leaks in v9 and again in v10, both of which were addressed by hotfixes.  Now I'm seeing the same type symptom on my EV11 servers.  Is anyone else seeing memory management issues on  the EV servers, primarily mailbox archiving and have you been able to pin down whether or not it is w3wp again?  This is the direction I'm headed now. 

Currrently we are seeing problemw wiht the archiving tasks running out of memory and the w3wp process's memory usage continually grows.  The current resolution is to reboot the server, although an IIS reset might also reset the memory pool as well.

Thoughts?

Best Practice
Enterprise Vault
Information Governance
Troubleshooting
Windows Server (2003-2008)
  • JesusWept3's avatar
    JesusWept3
    11 years ago

    have you ran a profiler against w3wp and the app pools to make sure that the memory never releases, and only increases regardless of connections? and also probably worth checking whether recycling the app pool releases any resources etc

     

    But you may be best off creating a case, my guess would be that its the EV Indexing as that is meant to be specc'd for a large memory footprint and its all http based as well, so you might see this being normal, but if its causing reboots and memory exhaustion , especially on an 8GB machine (which is the minimum spec) then would definitely contact symc

    You may want to get some full memory dumps either through debugdiag or ADPlus so that you have that info ready so they can determine whats happening with the memory allocation, also take a look at the HTTPERR logs to see if anything abnormal is happening with connections.

    Just as a matter of interest, what else is going on on the server? DA Searches? PST Migrations? Vault Cache builds? any IMAP connections?

    You may also want to look in to getting all your users Fast Browse enabled so the new Search (AE replacement) will query SQL and not the indexes, that may help a lot of the connections as well.

    And if you do install DebugDiag, make sure to deactivate any rules after you are finished, if the app pool is attached to a debuger, any issues will not be automatically resolved by IIS because its attached to a debugger etc

  • JesusWept3's avatar
    JesusWept3
    Level 6
    11 years ago

    have you ran a profiler against w3wp and the app pools to make sure that the memory never releases, and only increases regardless of connections? and also probably worth checking whether recycling the app pool releases any resources etc

     

    But you may be best off creating a case, my guess would be that its the EV Indexing as that is meant to be specc'd for a large memory footprint and its all http based as well, so you might see this being normal, but if its causing reboots and memory exhaustion , especially on an 8GB machine (which is the minimum spec) then would definitely contact symc

    You may want to get some full memory dumps either through debugdiag or ADPlus so that you have that info ready so they can determine whats happening with the memory allocation, also take a look at the HTTPERR logs to see if anything abnormal is happening with connections.

    Just as a matter of interest, what else is going on on the server? DA Searches? PST Migrations? Vault Cache builds? any IMAP connections?

    You may also want to look in to getting all your users Fast Browse enabled so the new Search (AE replacement) will query SQL and not the indexes, that may help a lot of the connections as well.

    And if you do install DebugDiag, make sure to deactivate any rules after you are finished, if the app pool is attached to a debuger, any issues will not be automatically resolved by IIS because its attached to a debugger etc

  • WiTSend's avatar
    WiTSend
    Level 6
    11 years ago

    Thanks for the suggestions and I'll be running through the various tests.  As far as what else is going on...  lots...  In addition to mailbox archiving of ~6,000 users per server we are also in the midst of an ArchiveShuttle migration of the archives moving off Centera and onto a new storage platform.  This activity is thus reindexing all the archives as we process them.  Also, at various times we have PST imports as part of aquisitions of various companies and their associated email. Other than that the server is pretty quiet...