See if the email error logs have anything in them.  If you are getting out
of memory errors, increase the initial/max heap size.  What is the last
message that was processed (incoming/outgoing)?

Axton Grams

On Fri, Sep 5, 2008 at 1:22 PM, strauss <[EMAIL PROTECTED]> wrote:

> We have been seeing the aremail engine hanging every few days
> (7.1.00.002 on Win2K3 Ent x64 w java 1.5.0_14 (32-bit) on 12gb RAM x 8
> core HP server) ever since we went live in May. With the fall semester,
> we have seen outbound mail traffic go up to 1200-1400 messages a day,
> and the crashes have increased in frequency. When the mail service
> crashes, it hangs in a state where there are over 1200 handles open,
> compared to where it usually sits below 1000 and runs up to 1300-1380
> when actually processing messages. Trying to stop it from the Services
> MMC results in a long wait while it tries to do so (it is hung, not
> dead), followed by a timeout error; then you can manually restart it.
>
> Support told us to upgrade the Outlook client from 2003 Sp1 to 2007,
> which we needed to do anyway since the mail box underneath it is
> switching over from Exchange 2000 to 2007.  That resulted in the service
> hanging almost hourly during the business day instead of once every day
> or two.  Most of the time when it hangs there is nothing in the
> stderr.log at all, but with Outlook 2007 it writes a 14kb detailed
> Exception Access Violation HotSpot Virtual Machine error log
> (hs_err_pid####.log) in the \Outlook12 directory (if it was doing this
> in the Outlook 11 directory we never saw it).  If you log into the
> server console after one of these events has happened, you get a
> deferred warning that the javaservice failed; clicking on the details
> gets you this:
>
> szAppName : aremaild.exe     szAppVer : 1.1.0.0     szModName :
> EMSMDB32.DLL
> szModVer : 12.0.4518.1014     offset : 000145b4
>
> or...
>
> szAppName : aremaild.exe     szAppVer : 1.1.0.0     szModName :
> ntdll.dll
> szModVer : 5.2.3790.3959     offset : 0005ec97
>
> In the event log you usually get:
>
> 12:49 PM: Faulting application aremaild.exe, version 1.1.0.0, faulting
> module ntdll.dll, version 5.2.3790.3959, fault address 0x0005ec97.
>  1:00 PM: Reporting queued error: faulting application aremaild.exe,
> version 1.1.0.0, faulting module ntdll.dll, version 5.2.3790.3959, fault
> address 0x0005ec97.
>  1:02 PM: Fault bucket 911867558.
>
> Last night I updated Outlook 2007 to Sp1 (it is still pointing at a 2000
> server mailbox), and added the line
> "External-Authentication-Return-Data-Capabilities: 31" to the ar.cfg and
> restarted the server. The email engine had been doing extensive lookups
> against our LDAP server for all kinds of notification information that
> isn't stored there, that is already in the User form, as evidenced in
> the arplugin.log, and since most of the service hangs occur during the
> processing of a group notification to 8-10 users in a single group, I
> thought that the delay imposed as it tried to fruitlessly look up data
> in LDAP might be causing the problem.  As it turns out, that was not the
> problem - the service has now hung six times in four hours with
> hs_err_pid####.log files, and appears to have restarted on its own a
> couple-three times, after which it does not always process mail. Several
> times when I have restarted the service, it has not even cleared a
> backlog of 50-100 messages before hanging again.
>
> Support has not been much help here, as we have had not one but two
> issues open since mid-June (they closed one issue in error and had to
> open another), so I wondered if anyone else had run into this severe a
> problem and found a solution that worked.  I am at the point of writing
> batch files and scheduling them to do a net stop (and wait for the hung
> service to release) followed by a net start every 15 minutes or so, and
> I have had to tell the entire IT staff here this morning that they
> cannot count on any of the email notifications to arrive in a timely
> manner.
>
> Why didn't they keep this engine in C, where it was rock solid for
> YEARS!???
>
> Christopher Strauss, Ph.D.
> Call Tracking Administration Manager
> University of North Texas Computing & IT Center
> http://itsm.unt.edu/
>
>
> _______________________________________________________________________________
> UNSUBSCRIBE or access ARSlist Archives at www.arslist.org
> Platinum Sponsor: www.rmsportal.com ARSlist: "Where the Answers Are"
>

_______________________________________________________________________________
UNSUBSCRIBE or access ARSlist Archives at www.arslist.org
Platinum Sponsor: www.rmsportal.com ARSlist: "Where the Answers Are"

Reply via email to