Host Monitor web interface fails after 1 hr, product limits

[davephan]

The Host Monitor web interface fails after one hour with the following error:

"Response from HostMonitor's Web Service:
Error: Cannot read data

Possible reasons:
- HostMonitor is not running
- RCI interface disabled
- HM Web Service configured incorrectly (wrong TCP Port or IP address)
- network problem"

To keep the Host Monitor web interace working, the following script is run once per hour with the Windows task scheduler to reset Host Monitor:

net stop "KS Web Service"
tasklist|find /i "hostmon.exe"
if errorlevel==0 Taskkill /IM hostmon.exe /F
sleep 30
net start "KS Advanced Host Monitor"
net start "KS Web Service"

Running the script does create a Host Monitor outage for several minutes while Host Monitor restarts. Another alternative might be to reboot the Host Monitor system once an hour instead of resetting the "KS Host Advance Host Monitor" and "KS Web Service" services.

When running Host Monitor as an application, instead of a service, the following error message frequently pops up, sometimes several times per minute:

"The script you are executing is taking longer than expected to run. Click End to abort the script or Continue to continue the script execution."

There are 264 systems being monitored with Host Monitor. There are a total of 8882 tests. When Host Monitor is started, the threads that hostmon.exe consumes builds up to about 1600 - 1800. The next highest thread counts used are svchost at 157 threads and system at 91 threads. When the thread count builds up to about 1600, then the Host Monitor
web interface fails. The Host Monitor web interface is then fixed by running a service reset script. The Host Monitor web interface then works for another hour before it fails again.

We seemed to reach a limit of the product when we added one more system. The Host Monitor web interface now fails after a few minutes. It only takes a few minutes for the thread count on the Host Monitor system to reach about 1600 threads.

I understand that Windows 32-bit systems have a thread count limit which is a bit higher than 1600. I suspect that Host Monitor is leaking threads, or cannot control the number of threads that are being used. The Windows 2003 32-bit server then exceeds it's limit of threads that can be run. The Host Monitor software was also tested on a Windows 2008 R2 64-bit server. The Host Monitor software crashed more frequently with the Windows 2008 R2 64-bit system.

I've tried drastically reducing the frequency of the tests, which seemed to slow down the thread build up time, but did not eliminate the problem. I also tried drastically reducing the number of tests, which also helped to reduce the thread build up time, but did not eliminate the problem. When starting the Host Monitor services, all the tests try to run right away. It seems that it would be better to have the initial test for each test start after each test's internal time, not try to all start right away. I've read that the product is limited to 128 tests per second. Does this mean staring
128 tests per second, or running 128 tests per second. If each test takes many seconds to run, then perhaps too many tests will be running at the same time. Is there a way to reduce the number of tests running per second? Many of the tests have intervals of 300 seconds. Some tests have long intervals, such as 900 - 1800 seconds.

What are the thread limits for Host Monitor? What are the test limits for Host Monitor? What are the limits for the number of systems being monitored?

Depending on the limits, maybe we need to have more than one Host Monitor server to cover all the tests, or drastically reduce the number of systems being monitored and/or the number and frequency of the tests.

Dave

[KS-Soft]

[KS-Soft]

What version of HostMonitor do you use?

Regards
Alex

[davephan]

We are using Advanced Host Monitor version 8.64 and are licensed for unlimited tests.

Could you be more specific which files need to be e-mailed?

Dave

[KS-Soft]

Unlimited tests? Do you mean Enterprise license?
We need HML file with tests (I don't know name of this file because you could use any name and path for the file) and all *.LST and *.INI files from HostMonitor directory.

Regards
Alex