Email send failure

[KS-Soft]

HostMonitor allows you to start as many copies as you want. It may warn you about another instance but it starts anyway. Of course you may start only 1 instance in service mode, but there is no limit for applications.
Just type hostmon.exe in command line on click on icon in program's group

Regards
Alex

[ericm]

What are some specs about your system you are running HM on
CPU ?
Mem?
HM Ver?
OS Ver?
.NET Framework Ver?
Other third party software or development software.

Reason I ask is because this error is not caused by HM in my opion. Even though HM is the reported program that is only because they system is running out of GDI resources.

goto http://www.sysinternals.com/Utilities/P ... lorer.html and download application
Install and Run
Click View and select colums
Click process memory tab and select GDI down at bottom

Then run HM and when the error generates you will be able to see what processes are using all your GDI.

Or
http://www.veign.com/download_app.asp?app=110
scroll to bottum of page and download. Once the application is started find host monitor in the program list highlight it click watches \ gdi \ set. then set a limit.
Do this for some other apps that ae usinging GDI also.
Then click on options and limit logging.
Then click tools start monitoring

Generate error and see what you get.

[AntonyP]

I have started an app instance oh HM, running on the same machine as the original service instance. I will monitor it for a couple of days and let you know of my findings.

[AntonyP]

Here are the results so far

After 4 hours of running HM as service (fully working) and as application (but with no alerts) i get these:

service uses 1% cpu - appl uses 10% cpu
service uses 60MB RAM - appl uses 10MB RAM
service uses 400 handles - appl uses 300 handles
service uses 50 threads - appl uses 15 threads
service uses 400 user objects - appl uses 200 user objects
service uses 6700 gdi objects - appl uses 300 gdi objects

[AntonyP]

Windows 2000 SP4
HM version 6.10

CPU P4 3.2GXZ
1GB RAM

[KS-Soft]

service uses 400 handles - appl uses 300 handles
service uses 50 threads - appl uses 15 threads
service uses 400 user objects - appl uses 200 user objects

I think such difference is normal. One instance started more tests then another at the moment when you have checked resource.

service uses 6700 gdi objects - appl uses 300 gdi objects

There is. So, we definately have resource leak in some action method. Good, we have reduced search area by 75%

Most tests use the email action (after 3 failed replies) plus an entry to the logfile of HM, which is a dbf one.

Any other actions? Could you please send all *.LST and *.INI files to support@ks-soft.net?

Regards
Alex

[AntonyP]

Just emailed them.

[KS-Soft]

Good news - we have reproduced this problem. Did not find mistake yet but I think we will find and fix it in couple days.
BTW: as we see GDI leak appears in service mode only, caused by "Send e-mail" action under some circumstances..

Does HostMonitor send a lot of e-mails in your environment? About 9 mails every minute?

Regards
Alex

[AntonyP]

The HM sends approximately 1 email per minute under "normal" circumstances, but there are cases when many tests fail and this causes the HM to send even up to 80 emails per minute. Let me clarify that if many tests fail silmutaniously, the HM has been configured in such way that is it practically impossible for it to send more than 80 emails max and only once, not every minute or according to the period of each failed test.

[AntonyP]

The huge resourse problem also disappears if i stop both alerts and monitoring of the HM and restart both, without stopping the whole service of HM.

[KS-Soft]

Apologize for delay. Problem located in compiler library and its very hard to fix the problem without some redesign of the entire library
We need some more time...
Temporary solution - start HostMonitor in application mode. No resource leak in such case, right?

Regards
Alex

[KS-Soft]

The huge resourse problem also disappears if i stop both alerts and monitoring of the HM and restart both, without stopping the whole service of HM.

You stop monitoring (using HostMonitor's menu) and HostMonitor releases GDI resources??
Are you sure?

Regards
Alex

[AntonyP]

The huge resourse problem also disappears if i stop both alerts and monitoring of the HM and restart both, without stopping the whole service of HM.

You stop monitoring (using HostMonitor's menu) and HostMonitor releases GDI resources??
Are you sure?

Regards
Alex

No, only the amount of RAM used is released, the rest remain the same

[AntonyP]

Apologize for delay. Problem located in compiler library and its very hard to fix the problem without some redesign of the entire library
We need some more time...
Temporary solution - start HostMonitor in application mode. No resource leak in such case, right?

Regards
Alex

It's ok with us, we know you are doing a fantastic job anyway

For the time being, i have set an alarm when HM uses too much RAM and if we (my colleagues and i) get such an alarm we just reload the HM service.

I hope it won't take you too much time to fix, hopefully less than a month ..

P.S. Just out of curiocity, i guess this kind of problem should have appeared with other customers of yours, ISP's etc., that also deal with large network monitoring. So, do you think it was a problem someone might have expected to come across eventually?

[KS-Soft]

only the amount of RAM used is released, the rest remain the same

RAM? Thats normal - HostMonitor (and Windows API) finishes started tests and actions and releases memory.

I hope it won't take you too much time to fix, hopefully less than a month

I hope so. We are working mostly on this problem.

Just out of curiocity, i guess this kind of problem should have appeared with other customers of yours, ISP's etc., that also deal with large network monitoring.

Yes, couple weeks ago we received e-mail... We could not reproduce the problem and symtoms were different. But I think that problems were caused by the same reason.

Regards
Alex