You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by geeky2 <ge...@hotmail.com> on 2012/06/28 22:49:31 UTC

avgTimePerRequest JMX M-Bean displays with NaN instead of 0 - when no activity

hello all,

environment: solr 3.5, jboss, wily

we have been setting up jmx monitoring for our solr installation.

while running tests - i noticed that of the 6 JMX M-Beans
(avgRequestsPerSecond, avgTimePerRequest, errors, requests, timeouts,
totalTime) ...

the avgTimePerRequest M-Bean was producing "NaN" when there was no search
activity.

all of the other M-Beans displayed a 0 (zero) when there was no search
activity.

we were able to compensate for this issue with custom scripting in wily on
our side.

can someone help me understand this inconsistency?

is this just a WAD (works as a designed) ?

thanks for any help or insight



--
View this message in context: http://lucene.472066.n3.nabble.com/avgTimePerRequest-JMX-M-Bean-displays-with-NaN-instead-of-0-when-no-activity-tp3991962.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: avgTimePerRequest JMX M-Bean displays with NaN instead of 0 - when no activity

Posted by Chris Hostetter <ho...@fucit.org>.
: while running tests - i noticed that of the 6 JMX M-Beans
: (avgRequestsPerSecond, avgTimePerRequest, errors, requests, timeouts,
: totalTime) ...
: 
: the avgTimePerRequest M-Bean was producing "NaN" when there was no search
: activity.
: 
: all of the other M-Beans displayed a 0 (zero) when there was no search
: activity.
	...
: is this just a WAD (works as a designed) ?

I believe it is ... avgTimePerRequest is simply the total amount of time 
spent processing requests divided by the total number of requests -- in 
your case "0/0" which is NaN.

All of the other metrics you mentioned are 0 when no requests have been 
processed because nothing about them results in division by zero.  (well, 
tehnically avgRequestsPerSecond could result in division by zero if you 
queried it at the exact millisecond that the handler was created, but 
that's not likely to ever happen).



-Hoss