up vote 1 down vote favorite
share [g+] share [fb]

This question relates to monitoring. The tool that I use (Check_MK) sets a number of threads warning limit at 2000. This number is taken from /proc/loadavg:

[foo@bar ~]$ cut /proc/loadavg -f4 -d\ 
 2/2001
[foon@bar ~]$

The number in question is the one after the forward-slash. Is it possible, given a number of CPUs (or other parameters) to calculate a threshold for this metric?

Thanks.

link|improve this question
feedback

1 Answer

up vote 1 down vote accepted

The value after the slash is the number of kernel scheduling entities that currently exist on the system

(source)

I am not sure how that is relevant for warning against too many threads. Also, value is dynamic, depending on system load and usage, so it's not possible to calculate it in advance.

link|improve this answer
Absolutely, it all depends on what the server is doing. However, I think there must be some kind of a ballpark, how many threads are too many? – ztron Feb 17 '11 at 18:11
I would start by monitoring server (for example email server) for week or so, and then take maximum number it reached. If everything was working fine then, it might be good idea to add some percent (for example 25%) to it, to prevent false alarms. As you said, it all depends on what the server is doing, AND how software is configured (is it starting huge number of threads or single-threaded). – Olli Feb 17 '11 at 18:23
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.