Hello list -
I've been trying to integrate the Radiator monitoring interface with Nagios and MRTG, specifically to graph the request rate and average response times on an ongoing basis. To do this, I'm using Net::Radiator::Monitor to query the Monitor port with StatisticsOnly and LogMicroseconds enabled. This lets me extract the variables "Average response time" and "Total requests"; the latter can be graphed as change over time to get request rate, and this works as expected. The problems I have are - * The 'Average response time' metric does not appear to be what it says - it is continually increasing, and is possibly total time spent handling responses? I cannot find any detailed documentation covering this. * All the statistics reset to zero when I do a reload or restart of Radiator. This would not normally be an issue, but we hold Node configuration in an externally managed database table, and this is apparently only read on restart or reload, so we need to regularly reload the config to obtain any updates to this table. Each time the statistics reset, I lose a polling cycle as it has to re-establish the interval increment. Can anyone clarify the meaning of the monitor parameters, specifically the Average response time one? Is there any way to make the Radiator software re-read the nodes table without resetting the statistics? All the other database tables (Yubikeys, TOTP keys etc) seem to be read on demand, but not the nodes table. Thanks in advance for any assistance. Steve Steve Shipway Unix Design Team Lead The University of Auckland T: +64 9 3737 599 ext 86487 E: s.ship...@auckland.ac.nz <mailto:s.ship...@auckland.ac.nz> (GNU Terry Pratchett)
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ radiator mailing list radiator@open.com.au http://www.open.com.au/mailman/listinfo/radiator
