Colin’s IT, Security and Working Life blog

December 17, 2009

Debugging Exchange 2010 W3WP High Memory Usage

Filed under: Fault Finding — chaplic @ 10:49 pm

 

I checked the Wordfish Exchange server and noticed high memory usage; in particular W3WP processes were consuming more memory than store.exe !

A quick Google produced nothing of note; the total memory usage was 1GB. On a server with two heavy users, that seems a tad high! One process was consuming more than 500MB alone.

TaskMan isn’t much help:

image

So, what next?

With headscratchers like this, Process Explorer is always a good bet.

Let’s fire it up, and look for our W3WP processes:

image

 

Let’s open the PID showing high memory usage to see if it gives us a clue:

image

Aha! This tells us the application pool!

Do we have a memory leak? Not sure, and without any web references we’re flying blind. Let’s jump into IIS, into application pools, and program some thread recycling for the thread:

image

 

Let’s recycle the thread now to confirm. Note it’s using 176MB of memory:

image

 

The replacement process is consuming less memory

 

image

Have I spotted a bug in Exchange IIS? Is this expected? Has my recycling helped? Don’t know yet. Only time till tell!

 

Technorati Tags: ,

:

 

 

:

About these ads

12 Comments »

  1. I am experiencing this as well and have yet to find any real answers! Please let me know what (if anything) you figure out!

    Comment by Keith — December 22, 2009 @ 9:44 pm

  2. Same here… any news about this leak?
    are you all using the OCS integration? can it be that?

    Comment by Julian — January 4, 2010 @ 2:43 pm

  3. Well, it’s been running now (with the regular recycling) since I posted that blog… and the highest w3wp is 145MB, the rest all under 100MB.

    There’s no OCS integreation here, though it does have the schema changes.

    My scientifically-unreliable conclusion: Memory leak.

    I’ll mention it to a MSFT contact to get his view as I’m not really wanting to raise a PSS case right now!

    Comment by chaplic — January 4, 2010 @ 5:10 pm

  4. Hello, any update?
    I have same problem.

    thx

    Comment by Voland_VVM — March 25, 2010 @ 7:55 am

  5. Having same problem here with a fresh 2008 R2 server and Exchange 2010 install. The max i’ve seen w3wp.exe go is about 1.5GB

    Did you ever find a solution?

    Comment by Justin — April 1, 2010 @ 6:42 am

  6. On close inspection, ours seems to be with “MSExchangePowerShellAppPool” instead of the OWA one. I’ve configured some recycling sessions in IIS and will see how we go …

    Comment by Justin — April 1, 2010 @ 6:53 am

  7. Exactly the same issue here with 2010 sp1 install and server with latest recommended updates. 100 mailboxes, 16GB of RAM, OWA and Blackberry users and a msExchESEparamCacheMaxSize set to 8GB.

    After couple of hours the RPC or OWA connection are terribly slow,
    Killing some of w3wp.exe process solved immediatly the latency.

    Performing a recycling every 360mn as a workaround worked. Looking foreward to a Hotfix!

    Comment by yannick — October 12, 2010 @ 2:21 pm

  8. Thanks for documenting this process.
    I had the same situation as Justin with powershell being the culprit.
    The powershell application pool consumed approximately 500MB.
    Scheduled a daily recycling event to refresh. process explorer is an invaluable tool found here:
    http://download.sysinternals.com/Files/ProcessExplorer.zip

    http://technet.microsoft.com/en-us/sysinternals/bb896653.aspx

    Comment by JohnD — November 4, 2010 @ 1:57 pm

  9. I read the following comment on a forum: “Most of that memory usage is for cache. Exchange will cache data to avoid using a more costly way to access it (disk read or network read). It is ok if your server uses all of it’s available memory if the other systems are not over utilized.”

    I would largely agree with that but in the case of our current 2010 implementation (on Hyper-V) becomes incredibly sluggish (tools and RDP response).

    Comment by Nadim J — March 13, 2011 @ 1:08 pm

  10. sweet jesus, this saved my day! I am going to schedule this for every morning! can be sooo annoying!

    Comment by Mick — January 10, 2012 @ 2:41 am

  11. to update reply 7, our error was our DC’s did not have AV exlusions configured. It was ok for Exchange 2003 talking to the DC’s for the last 5 years but not when implementing Exchange 2010. 2010 is a lot more chatty that 2003 and our AV was taking it as a threat. I got the issue on 2 differents implementation.

    On top make sure if you have remote sites, your DC are configured correctly in AD site and services to avoid Exchange talking to the remote site GC’s.

    Comment by yannick — February 6, 2012 @ 9:56 pm

  12. Hello Exchange users, Symantec Endpoint protection was my culprit. Eventhough it was disabled the SMC services were inhibiting connections to my server. At first thought I was considering RPC issues, however now that it was completely uninstalled my server is completely restablized.

    Comment by Rich — March 15, 2012 @ 4:57 pm


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

The Silver is the New Black Theme Blog at WordPress.com.

Follow

Get every new post delivered to your Inbox.

%d bloggers like this: