View previous topic :: View next topic |
Author |
Message |
leo1631
Joined: 27 Apr 2006 Posts: 55
|
Posted: Thu Nov 09, 2006 7:56 pm Post subject: A bug I think, HELP! |
|
|
Advanced Host Monitor 6.24 Enterprise edition + Lifetime updates.
A http test, which internal is 5 min, internal will change to 1 min if bad.
This test's info pane shows like this:
...
total tests 212
passed 211(100%)
failed 1(0%)
unknown 0(0%)
total time 17:41:05
alive time 17:35:46(99%)
dead time 00:05:18(1%)
...
This test's private log(full) shows like this:
...
2006-11-8 9:32:54 host is alive
2006-11-8 9:38:12 no answer
2006-11-8 9:39:23 host is alive
...
---------------
The PROBLEM is, I think the "dead time" in info pane should be "00:01:11", not "00:05:18", cause I change the internal from 5 min to 1 min when the test is bad, private log can prove that.
I think it is a bug, please check that.
I've autogenerated reports in HostMonitor(not log analyzer) every month, and especially shown them to my boss. The inaccurate "dead ratio" and "dead time" are deadly.
I'm crazy nearly, Please help me as soon as possible, thanks a lot. |
|
Back to top |
|
|
KS-Soft
Joined: 03 Apr 2002 Posts: 12806 Location: USA
|
Posted: Fri Nov 10, 2006 12:17 am Post subject: |
|
|
Quote: |
2006-11-8 9:32:54 host is alive
2006-11-8 9:38:12 no answer
2006-11-8 9:39:23 host is alive
The PROBLEM is, I think the "dead time" in info pane should be "00:01:11", not "00:05:18", cause I change the internal from 5 min to 1 min when the test is bad, private log can prove that |
9:38:12 - 9:32:54 = 00:05:18
9:39:23 - 9:38:12 = 00:01:11
Which "dead" time is more correct? We cannot say. Because we know server status at 9:32 and 9:38 and 9:39 but we don't know status of the system between probes. May be real dead time is 9:39:23-9:32:54 = 00:06:29? or 00:02:20? or 00:00:02?
So its impossible to provide 100% accurate report (unless you check server every second). HostMonitor takes 1st interval. Its not a bug. If we change behaviour and HostMonitor will take 2nd interval it does not change anything. Somebody else will complain.
Regards
Alex |
|
Back to top |
|
|
leo1631
Joined: 27 Apr 2006 Posts: 55
|
Posted: Fri Nov 10, 2006 3:08 am Post subject: |
|
|
Understand, but I insist my method.
Thanks anyway. |
|
Back to top |
|
|
KS-Soft
Joined: 03 Apr 2002 Posts: 12806 Location: USA
|
Posted: Fri Nov 10, 2006 12:20 pm Post subject: |
|
|
Your method is not better then ours.
I can offer more correct method for average statistics: BadTime = (Time3-Time1)/2 = (9:39:23-9:32:54)/2 = 00:03:15
Regards
Alex |
|
Back to top |
|
|
leo1631
Joined: 27 Apr 2006 Posts: 55
|
Posted: Sun Nov 12, 2006 4:20 am Post subject: |
|
|
You are correct, do that if possible. |
|
Back to top |
|
|
KS-Soft
Joined: 03 Apr 2002 Posts: 12806 Location: USA
|
Posted: Sun Nov 12, 2006 5:02 pm Post subject: |
|
|
Ok, I have added your request into "to do" list
Does anyone agree with such modification?
Regards
Alex |
|
Back to top |
|
|
AntonyP
Joined: 02 Jan 2006 Posts: 159 Location: Athens Greece
|
Posted: Mon Nov 13, 2006 1:11 am Post subject: |
|
|
2006-11-8 9:32:54 host is alive
2006-11-8 9:38:12 no answer
2006-11-8 9:39:23 host is alive
Atmo, dead time should be 9:32:54-9:39:23. The best way to avoid "2006-11-8 9:38:12 no answer" is to set an extra check (if no response then recheck xxx times). At least that is what i have configured and it works great! |
|
Back to top |
|
|
KS-Soft
Joined: 03 Apr 2002 Posts: 12806 Location: USA
|
Posted: Mon Nov 13, 2006 3:27 pm Post subject: |
|
|
So many people, so many tastes
May be we should keep it the way it works for last 5 years? So, nobody will be surprised...
Regards
Alex |
|
Back to top |
|
|
AntonyP
Joined: 02 Jan 2006 Posts: 159 Location: Athens Greece
|
Posted: Tue Nov 14, 2006 1:29 am Post subject: |
|
|
KS-Soft wrote: | So many people, so many tastes
May be we should keep it the way it works for last 5 years? So, nobody will be surprised...
Regards
Alex |
I agree |
|
Back to top |
|
|
Steven
Joined: 16 Feb 2006 Posts: 44
|
Posted: Wed Nov 15, 2006 12:36 pm Post subject: |
|
|
I understand how it's impossible to be 100% accurate here. But I do like the average idea. And it doesn't seem like a lot of work to implement either. The only problem is... explaining this to ALL THE USERS who have been using this for the past 5 years.. |
|
Back to top |
|
|
AntonyP
Joined: 02 Jan 2006 Posts: 159 Location: Athens Greece
|
Posted: Sat Nov 18, 2006 11:12 am Post subject: Re: A bug I think, HELP! |
|
|
leo1631 wrote: | Advanced Host Monitor 6.24 Enterprise edition + Lifetime updates.
A http test, which internal is 5 min, internal will change to 1 min if bad.
This test's info pane shows like this:
...
total tests 212
passed 211(100%)
failed 1(0%)
unknown 0(0%)
total time 17:41:05
alive time 17:35:46(99%)
dead time 00:05:18(1%)
...
This test's private log(full) shows like this:
...
2006-11-8 9:32:54 host is alive
2006-11-8 9:38:12 no answer
2006-11-8 9:39:23 host is alive
...
---------------
The PROBLEM is, I think the "dead time" in info pane should be "00:01:11", not "00:05:18", cause I change the internal from 5 min to 1 min when the test is bad, private log can prove that. |
Well if HM support wants to change the default settings, i guess i would prefer the above, as leo1631 suggested. |
|
Back to top |
|
|
KS-Soft
Joined: 03 Apr 2002 Posts: 12806 Location: USA
|
Posted: Fri Dec 08, 2006 4:44 pm Post subject: |
|
|
Quote: | I can offer more correct method for average statistics: BadTime = (Time3-Time1)/2 = (9:39:23-9:32:54)/2 = 00:03:15 |
Done. Version 6.50 Beta available at http://www.ks-soft.net/hostmon.eng/downpage.htm
However this is undocumented option. if you want to change statistics calculations, add UseBothChecksForStat=1 line into [Misc] section of hostmon.ini file and restart HostMonitor
Regards
Alex |
|
Back to top |
|
|
|