Re: [Nagios-devel] How is latency calculated?
Posted: Mon Mar 30, 2009 3:28 pm
If I don't load the module at all I get latencies in the 10,000 range. I'v=
e not been able to graph them to see over a very long time, but as it's fly=
ing off the chart it just looks like a sharp upward spike.
My guess is 27,000 service checks are just too many for a single nagios ins=
tance to run.=20
________________________________________
From: Andreas Ericsson [[email protected]]
Sent: Monday, March 30, 2009 9:56 AM
To: Nagios Developers List
Subject: Re: [Nagios-devel] How is latency calculated?
Steven D. Morrey wrote:
> Hello Everyone,
>
> I've been working on updating a Neb Module and bringing it up to compatib=
ility with Nagios 2.12 with an eye towards getting updated to 3x.
> I've run into a bit of a snag though.
>
> It appears that service check latency climbs along a steady and continual=
curve. This curve can be exaggerated by having the module immediately ret=
urn with fake results.
> For instance under normal circumstances service check latency starts out =
at about 30 seconds (there are some 27,000 checks being run), this latency =
will continually grow in a relatively smooth curve at a rate of about 2 sec=
s per hr.
> However if I modify the Neb module to immediately return ok, instead of d=
ecreasing service check latency it actually increases it, so instead of 2 s=
ecs per hr of growth, it skyrockets to 200 seconds per hr.
>
> I just can't see how this could be possible. One would tend to think tha=
t having it return immediately, thereby short circuiting the check, that la=
tency should drop to around 0.
> Instead it appears that latency grows as the number of checks completed i=
ncreases.
>
That is indeed weird. What happens if you don't load the module at all? Sky=
rocket or
slow-but-steady increase?
> To top it off, I've grepped the source numerous times to try and find the=
service check latency calculation and I'm not seeing anything, at least no=
t in the reaper which is where I would expect to find that calculation.
>
> Can anyone give me some advice here? Especially if I could know where th=
e latency, both for the individual service check itself as well as average =
service check latency for the system is being calculated, I would very much=
appreciate it.
I'll look it up tomorrow. Office-time is officially over in sweden now,
and I'm heading home.
--
Andreas Ericsson [email protected]
OP5 AB www.op5.se
Tel: +46 8-230225 Fax: +46 8-230231
Considering the successes of the wars on alcohol, poverty, drugs and
terror, I think we should give some serious thought to declaring war
on peace.
---------------------------------------------------------------------------=
---
_______________________________________________
Nagios-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/lis ... gios-devel
NOTICE: This email message is for the sole use of the intended recipient(s=
) and may contain confidential and privileged information. Any unauthorized=
review, use, disclosure or distribution is prohibited. If you are not the =
intended recipient, please contact the sender by reply email and destroy al=
l copies of the original message.
This post was automatically imported from historical nagios-devel mailing list archives
Original poster: ndreas Ericsson [[email protected]
e not been able to graph them to see over a very long time, but as it's fly=
ing off the chart it just looks like a sharp upward spike.
My guess is 27,000 service checks are just too many for a single nagios ins=
tance to run.=20
________________________________________
From: Andreas Ericsson [[email protected]]
Sent: Monday, March 30, 2009 9:56 AM
To: Nagios Developers List
Subject: Re: [Nagios-devel] How is latency calculated?
Steven D. Morrey wrote:
> Hello Everyone,
>
> I've been working on updating a Neb Module and bringing it up to compatib=
ility with Nagios 2.12 with an eye towards getting updated to 3x.
> I've run into a bit of a snag though.
>
> It appears that service check latency climbs along a steady and continual=
curve. This curve can be exaggerated by having the module immediately ret=
urn with fake results.
> For instance under normal circumstances service check latency starts out =
at about 30 seconds (there are some 27,000 checks being run), this latency =
will continually grow in a relatively smooth curve at a rate of about 2 sec=
s per hr.
> However if I modify the Neb module to immediately return ok, instead of d=
ecreasing service check latency it actually increases it, so instead of 2 s=
ecs per hr of growth, it skyrockets to 200 seconds per hr.
>
> I just can't see how this could be possible. One would tend to think tha=
t having it return immediately, thereby short circuiting the check, that la=
tency should drop to around 0.
> Instead it appears that latency grows as the number of checks completed i=
ncreases.
>
That is indeed weird. What happens if you don't load the module at all? Sky=
rocket or
slow-but-steady increase?
> To top it off, I've grepped the source numerous times to try and find the=
service check latency calculation and I'm not seeing anything, at least no=
t in the reaper which is where I would expect to find that calculation.
>
> Can anyone give me some advice here? Especially if I could know where th=
e latency, both for the individual service check itself as well as average =
service check latency for the system is being calculated, I would very much=
appreciate it.
I'll look it up tomorrow. Office-time is officially over in sweden now,
and I'm heading home.
--
Andreas Ericsson [email protected]
OP5 AB www.op5.se
Tel: +46 8-230225 Fax: +46 8-230231
Considering the successes of the wars on alcohol, poverty, drugs and
terror, I think we should give some serious thought to declaring war
on peace.
---------------------------------------------------------------------------=
---
_______________________________________________
Nagios-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/lis ... gios-devel
NOTICE: This email message is for the sole use of the intended recipient(s=
) and may contain confidential and privileged information. Any unauthorized=
review, use, disclosure or distribution is prohibited. If you are not the =
intended recipient, please contact the sender by reply email and destroy al=
l copies of the original message.
This post was automatically imported from historical nagios-devel mailing list archives
Original poster: ndreas Ericsson [[email protected]