RCC unstable when updating test list

CapQwerty · Joined: 14 Dec 2005 Posts: 36

Hi,

We are having some problems with the RCC being unstable when we add, delete or modify tests. The interface will freeze for several seconds while saving the changes, and often the client will lose its connection to the HostMonitor server and have to reconnect. This happens particularly if we update more than one test at a time.

Our test list has grown a lot over the last few months, which is when we started seeing this problem. I'm not sure if that might have anything to do with it, but we now have ~2000 tests.

We are using:
- RCC v2.44
- HostMonitor v7.50 (running in service mode)
- Windows Server 2003

Are you able to suggest what the problem might be or anything we can do to fix it?

Thanks,
CapQwerty

KS-Soft · Joined: 03 Apr 2002 Posts: 12795 Location: USA

For begining, lets update RCC. HostMonitor 7.50 comes with RCC 2.56.
Its a bad idea to mix components from different versions of the software.

Regards
Alex

CapQwerty · Joined: 14 Dec 2005 Posts: 36

Hi Alex,

Thanks, I didn't realise my RCC was out of date. Unfortunately, the new version still has the same problem.

Here's an example:

I select about 10 tests, right click and disable them.

RCC freezes for about 20 seconds, then I get a popup saying "Warning. Connection lost. Try to reconnect?"

If I select yes, RCC freezes again for about 20 seconds, then I get another popup saying "Error. Cannot read dataH", or just a blank popup, or sometimes some garbage characters.

If I close RCC and reconnect, everything looks OK and the tests are disabled. If I try to enable them again, the process is pretty much the same.

I notice that while the RCC is frozen after I initially modify the tests, the HostMonitor executable running on the server seems to be using 100% CPU on one core. At about the same time I get disconnected, the CPU drops back to normal. The RCC itself isn't using any CPU at all.

Thanks,
CapQwerty

KS-Soft · Joined: 03 Apr 2002 Posts: 12795 Location: USA

I think software drops connection because HostMonitor does not respond within specified timeout. You may increase specified timeout interval as workaround.
Question is: why HostMonitor does not respond for 20 sec? Could you send your configuration to support@ks-soft.net? We need HML file with tests plus all *.INI and *.LST files.

Regards
Alex

CapQwerty · Joined: 14 Dec 2005 Posts: 36

Hi Alex,

We have emailed the configuration files as requested.

Thanks,
CapQwerty

KS-Soft · Joined: 03 Apr 2002 Posts: 12795 Location: USA

Thank you for the files.
Some results of testing:
1) We could not reproduce "connection lost" problem. Probably because you have increased timeout?
2) HostMonitor uses significant CPU resources because it saves entire test list after any config modification. You are using "Auto save testlist after any changes" option (Preferences page in the Options dialog).
HostMonitor uses below 30% of CPU on our system for 3 seconds. I think it can be a problem if you have old and slow system or you have instaled some antivirus monitor that checks file modifications in real-time.
You may try to change that option to "Save tests and stats at midnight" and setup your antivirus monitor (if you have any) to skip HML file with tests.

Also, we will optimize our code a little. We can reduce system load in case you modify common test properties for several items at the same time.

Regards
Alex

KS-Soft · Joined: 03 Apr 2002 Posts: 12795 Location: USA

PS
There is optimized version: www.ks-soft.net/download/hm752.zip

Regards
Alex

CapQwerty · Joined: 14 Dec 2005 Posts: 36

Hi Alex,

Thanks for the feedback.

The connection lost problem only happens when we are running the RCC from another machine - if we run it locally on the HostMonitor server then the screen still freezes but the connection doesn't break.

Here are our server specs, I think they should be more than sufficient for the task:

Windows Server 2003 SP2
2 x Dual Core Intel Xeon 5140 CPUs @ 2.33GHz
6GB RAM

We have tried both excluding the HostMonitor files from virus scanning and disabling our antivirus software completely, but the problem still occurs.

The optimised version seems to behave exactly the same as the normal version - it still takes about 20 seconds to save and times out.

Changing the saving behaviour made a massive difference to the performance. When we switch to "Save tests and stats at midnight" we can disable / enable / modify large numbers of tests easily, with no freezing at all.

What is strange is that even when we make changes to large numbers of tests and then click "Save" manually, the save only takes a few seconds and we don't get disconnected.

Is it possible that in the "Auto save TestList after any changes" mode, when we make changes to multiple tests at the same time HostMonitor is trying to save the test list after updating every test? So if we selected 10 tests and clicked disable, would it disable one, save, then disable another, save, and so on? That would explain the very poor performance of the autosave mode.

If we keep running in the "Save tests and stats at midnight" mode, is there any way that we can trigger HostMonitor to save the test list automatically at other times? Are there any commands we can issue using the telnet service or any other means that will cause the test list to be saved? We are a bit concerned about losing a whole day's worth of changes if the machine or app were to crash without saving.

Thanks,
CapQwerty

KS-Soft · Joined: 03 Apr 2002 Posts: 12795 Location: USA