Re: Losing NT Services

Question

Tran,
Forte is a powerful tool which, while it protects you from some issues,
does not fully insulate you from writing code that can peg out a
machine.
So, for example, it would be a simple 10 line or so Forte program which
could issue 'start task' statements inside an infinite loop which would
then soon max out the available threads on a machine. I could write
that same program in C, perl, or any other environment. I'm not saying
that you wrote such a program, just that by writing it in Forte doesn't
mean that "forte" is degrading the system - the program is degrading the
machine.
Forte does not allow you to ignore and overcome finite resources on a
machine (i-nodes, threads, processes, memory, disk space, etc). And
this problem sounds like its related to the consumption of system
resources.
I'm not just pointing the finger back at you, rather I'm suggesting that
for the forum to be able to help you it is important for you to describe
some more detail about what the forte system is doing. Is it a
development server? an application Server? How many partitions are
running on the box, and most importantly, what are they trying to do?
These are the same questions that would be asked of you if you wrote
your application in C++ or powerbuilder (gasp!) or whatever.
Is the file transfer a part of your application's functionality?
Have you tried running the performance monitor to watch things like
available memory, disk activity, process count, etc?
We have been working with Forte on NT for some time now, and while there
are issues, what you describe has not been our general experience. We
have, however, managed to peg out NT machines in specific situations.
-John
We are hoping that someone have ran into the same problem and can shed
some
light on any work around to resolve it. The problem that we are
encountering is that the NT Server's performance degrade to a point
where
we physically can't sign on to the box itself. When someone sits at the
terminal and tries to sign on, the person would see a blue screen and
that's it. The only thing that is running on that box is Forte'
runtime,
pcAnyWhere, and SQL Server. Once the NT server get to this point, all
we
can do is reboot the box. The key indicator that NT is dying is our
application will start to malfunction and various files transfer would
fail.
Does anyone have an application running with NT, if you do we would love
to
have a conversation concerning this situation. We would appreciate any
advice on this matter.
John Jamison [email protected]
Vice President and Chief Technology Officer
Sage IT Partners, Inc.
Voice: 415 392-7243 x 306
Fax: [phone]
Internet Enabled Business Change
http://www.sageit.com
-----------------------------------------------------

Answer 1

Hi-
I don't know if this will shed any light on your problem, but I can relate to you one circumstance that we had on our server in the development environment. When you have a process such as a service object that crashes, it will sometimes leave an ftexec hanging. Each of these "dead" ftexec's seem to tie up a certain amount of memory and will not let go. If something like that was going on and went unchecked, I could well imagine that it would eventually bring NT to its knees.
If you are running the forte environment as an NT service, you can see this by running the task manager. (ctrl+alt+delete, then click the task manager button, then select the processes tab.) If you see a lot of ftexec's listed, you may have something similar occurring. Even though the Task Manager may report little or no memory allocated to them, somehow it is tied up somewhere. For grins, try this: Bring up the task manager, and then stop the Forte environment. Watch what happens to your memory utilization as each of the dead processes goes away.
Unfortunately, if this is your problem, I do not have a solution for you, but maybe someone else does. This was one of those things that I made a mental note that I need to figure out someday, but it was not causing us a serious problem, so the resolution was not a high priority. We were training several new developers at the time, and they were fairly green at both Forte and SQL, so I think they had a lot of dbsession objects crashing. In our case, it was sufficient to restart the Forte environment if performance started to degrade. And as the developers got better at SQL and at catching exceptions, we had fewer instances of dead objects to deal with.
However, if anyone knows how to resolve or prevent this problem, I would certainly be interested in hearing the solution.
-----Original Message-----
From: Thanh T. Tran [SMTP:[email protected]]
Sent: Monday, May 25, 1998 11:28 PM
To: [email protected]
Subject: Losing NT Services
We are hoping that someone have ran into the same problem and can shed some
light on any work around to resolve it. The problem that we are
encountering is that the NT Server's performance degrade to a point where
we physically can't sign on to the box itself. When someone sits at the
terminal and tries to sign on, the person would see a blue screen and
that's it. The only thing that is running on that box is Forte' runtime,
pcAnyWhere, and SQL Server. Once the NT server get to this point, all we
can do is reboot the box. The key indicator that NT is dying is our
application will start to malfunction and various files transfer would fail.
Does anyone have an application running with NT, if you do we would love to
have a conversation concerning this situation. We would appreciate any
advice on this matter.
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>