I think it would help, if you showed how you are measuring queries per second.
Off the top of my head ntpdc sysstats, or ntpdc ifstats or ntpdc iostats? mrtg and many other traffic graphers know that counters overflow or may be randomly reset by power cycles or other administrative things, and have some simple logic to throw outliers out. This isn't so much a "buffer overflow" as it is "counter deltas must be physically possible" (a powerful technique in any sort of data acquisition based on a physical understanding of what you're trying to measure). These are kind of automatic in mrtg unless you take pains to disable them. Don't know how to configure these in munin. ________________________________________ From: [email protected] [[email protected]] On Behalf Of Jörg Hoffmann [[email protected]] Sent: Sunday, November 21, 2010 7:08 AM To: [email protected] Subject: [Pool] mysterious munin stats Hallo Everybody, I’ve running Munin for a very long time and I’ve noticed this bigger getting spikes every day. I´m not sure if it’s a buffer overflow but it seems so be something like it. http://netstat.root24.eu/munin/Root24/Root24/ntp.html the ntpdc is 64bit. Maybe someone has an idea what happens then or how to get “better” stats. http://www.pool.ntp.org/scores/91.121.20.142 Greetings Jörg _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
