Intermittent global ping test fails

KS-Soft · Post by **KS-Soft** » Mon Mar 02, 2015 11:33 am

This means you have several hundrend master test. 200? 600? 900?
In such case 5 sec may be too short interval.

If you have 200 master tests, you can keep 5 sec for masters but increase "Don't start more than N tests per second" option. Set at least 64

If you have 600 master tests, I would increase interval to 15 and set "Don't start more than N tests per second" to 64 as well.

Regards
Alex

ckratsch · Post by **ckratsch** » Mon Mar 02, 2015 12:12 pm

All right. I've been fiddling with the Behavior settings, and it looks like when I have "Don't start more than N tests per second" set higher than 6, CPU usage (Xeon 2.33Ghz dual core) jumps to around 50%, and the errant condition recurs.

I'll leave it at six for now, and revisit moving Hostmon to a new piece of hardware.

Thanks again.

KS-Soft · Post by **KS-Soft** » Mon Mar 02, 2015 12:48 pm

6 leads to high CPU Usage? That's strange. On "normal" system you may run 60 tests per second without stress.

What other test methods do you use?
ODBC logging? ODBC test methods?
May we see Auditing Tool screen shot?

May be a lot of log files? or one big log file? Antivirus?
Antivirus may check files when HostMonitor updates logs and leads to high CPU usage

>I'll leave it at six for now

With average load 7 tests per second? Plus 600 master tests?
This means HostMonitor will not be able to perform all tests.

Regards
Alex

ckratsch · Post by **ckratsch** » Mon Mar 02, 2015 2:00 pm

Most of our tests are ping tests, but we also use these Windows tests:

Drive space check
Service check
Memory usage check
CPU usage check
Event log check

We're not doing anything ODBC with Hostmon.

Our log.htm and syslog.htm files were ~2GB. I just recycled those.

The Hostmonitor directory was already excluded from AV scanning; I disabled all realtime scanning for testing. No change.

KS-Soft · Post by **KS-Soft** » Mon Mar 02, 2015 2:53 pm

23 Shell Scripts per min. Some heavy scripts? May be this is the problem?
Could you try to disable these items?

If this will not help, could you send config files to us (*.HML, *.LST, *.INI files to support@ks-soft.net)?
If there is some config/HostMonitor error, we should be able to reproduce the problem.
Otherwise I would suggest to move HostMonitor to different system.

Regards
Alex

ckratsch · Post by **ckratsch** » Mon Mar 02, 2015 2:55 pm

I'll work on that. Those shell scripts are small, used for % free memory checking (before Hostmon had it natively).

ckratsch · Post by **ckratsch** » Mon Mar 02, 2015 4:21 pm

Removed all the shell script tests, didn't help.

I decided to grab Process Explorer and sort out what exactly is launching the CPU usage up. It's svchost.exe, and I've narrowed it down to being the Network Store Interface Service. Sadly, this is a required service for network connectivity.

As before, I realize this may be out of your scope, and we're also looking into hardware replacement. I'll get those config files together so you can look at them.

ckratsch · Post by **ckratsch** » Wed Mar 04, 2015 2:48 pm

Tentative good news. After updating all drivers and firmware with the Dell Server Update Utility, I am no longer getting these symptoms, even if I bump concurrent tests to 128.

Sorry for the bother!

KS-Soft · Post by **KS-Soft** » Wed Mar 04, 2015 3:04 pm

Good news.
We tested your config as well and did not see any problems (beside too low threads limit).

Regards
Alex