No. Only have single processor machines.
One thing we have done with our "tmBoth" components is to aggregrate the COM Free
Threaded Marshaler.
We did find some interesting bugs, but they were of our own making. One thing was that
for some "safecalls", exceptions were not getting trapped. We added real try ..
except blocks which did trap the AV's and then managed to debug it from there. Also
added plenty of Asserts() and logged problems to the NT event log which helped.
Myles.
-----Original Message-----
From: Ramanlal, Vikash [SMTP:[EMAIL PROTECTED]]
Sent: Thursday, May 06, 1999 9:26 AM
To: Multiple recipients of list delphi
Subject: RE: [DUG]: COM Stress Testing.
All the threads in our test program are created by the main VCL thread and
is safe. Also in real life these COM objects are created by ASP pages so
that should be safe.
Are you running on a multi-processor NT server. This is where we see our
problem.
Regards
Vikash
> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]
> Sent: Thursday, 6 May 1999 08:34
> To: Multiple recipients of list delphi
> Subject: RE: [DUG]: COM Stress Testing.
>
> Did you also take make the Delphi thread creation thread safe - as
> suggested by Maurice Butler?
>
> I have been testing our component okay with upto 80 threads with no
> problems now occur. These are both a mixture of STA and MTA (Single
> threaded apartments and Multi-Threaded apartments).
>
> Myles.
>
>
> -----Original Message-----
> From: Ramanlal, Vikash [SMTP:[EMAIL PROTECTED]]
> Sent: Wednesday, May 05, 1999 11:25 AM
> To: Multiple recipients of list delphi
> Subject: RE: [DUG]: COM Stress Testing.
>
> We have written several COM objects in Delphi4 (Threading Model of
> Both) and
> running on a multi processor NT found that we got access violations
> and
> invalid pointer operations.
>
> To isolate the problem, we ended up creating a simple COM object
> (both
> threading model) and a test program that creates and destroys this
> object in
> several threads for a specified number of iterations. We got access
> voilations in the in-process COM server DLL when we used only 2
> threads.
> Now this object is just the shell created by the D4 wizard. So one
> would
> have expected it to work okay.
>
> Any ideas what could be wrong?
>
> My suspicion is that the ComServer object is not completely thread
> safe and
> on multi processor systems we are seeing a race condition.
>
> I put a critical section in the creation of the COM objects in the
> Client
> App and presto, it works!.
>
> The only problem is the the client App our COM objects will be used
> from is
> IIS/ASP. We cannot place such as locking mechanism here.
>
> Any Ideas?
>
>
> Regards
> Vikash
>
> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, 5 May 1999 10:26
> To: Multiple recipients of list delphi
> Subject: RE: [DUG]: COM Stress Testing.
>
>
>
> Server dosn't have the same limitation.
>
>
>
>
>
>
>
> Myles Penlington <[EMAIL PROTECTED]> on 05/05/99 09:26:54
>
> Please respond to [EMAIL PROTECTED]
>
> To: Multiple recipients of list delphi <[EMAIL PROTECTED]>
> cc: (bcc: Peter Jones/Logistics&Information
> Technology/Christchurch/Foodstuffs)
> Subject: RE: [DUG]: COM Stress Testing.
>
>
>
>
> Yep NT4.0 Workstation - Does this mean it will not occur on NT
> server??
> It's not DCOM, I'm just running a COM component and the test program
> on the
> same machine.
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [SMTP:[EMAIL PROTECTED]]
> Sent: Tuesday, May 04, 1999 2:45 PM
> To: Multiple recipients of list delphi
> Subject: Re: [DUG]: COM Stress Testing.
>
>
>
> You don't happen to be using NT Workstation for your server do you?
> 10
> Connections is the magic number for Workstation.
>
>
>
>
>
>
> Myles Penlington <[EMAIL PROTECTED]> on 04/05/99 14:58:32
>
> Please respond to [EMAIL PROTECTED]
>
> To: Multiple recipients of list delphi <[EMAIL PROTECTED]>
> cc: (bcc: Peter Jones/Logistics&Information
> Technology/Christchurch/Foodstuffs)
> Subject: [DUG]: COM Stress Testing.
>
>
>
>
> I have written a D4 program to stress test our D4 COM object.
>
> All works fine up until about 10 threads, after which we can
> guarantee to
> get a "OLE Catastrophic error". Does anybody know if this is a MS
> COM
> problem or a Delphi problem?
>
> Does anybody know what causes the "Catastrophic error"??
>
> Regards
>
> Myles Penlington
> Software Engineer
> ADVANCED MANAGEMENT SYSTEMS
>
> This message is confidential to the person to whom it is addressed.
> It is
> not to be used, copied or distributed by anyone else. If the
> message has
> reached you by mistake please contact us immediately and then delete
> it
> from your system. Thank you for your co-operation.
>
>
>
> --------------------------------------------------------------------------
> -
> New Zealand Delphi Users group - Delphi List -
> [EMAIL PROTECTED]
> Website: http://www.delphi.org.nz
>
>
>
>
>
>
>
> --------------------------------------------------------------------------
> -
> New Zealand Delphi Users group - Delphi List -
> [EMAIL PROTECTED]
> Website: http://www.delphi.org.nz
>
>
> (UUEncoded file named: att1.unk follows)
> (Its format is: File type unknown )
>
>
>
>
> --------------------------------------------------------------------------
> -
> New Zealand Delphi Users group - Delphi List -
> [EMAIL PROTECTED]
> Website: http://www.delphi.org.nz
>
> --------------------------------------------------------------------------
> -
> New Zealand Delphi Users group - Delphi List -
> [EMAIL PROTECTED]
> Website: http://www.delphi.org.nz
---------------------------------------------------------------------------
New Zealand Delphi Users group - Delphi List - [EMAIL PROTECTED]
Website: http://www.delphi.org.nz
application/ms-tnef
