Thanks Alex. I had actually just been pushing zeros into the DB while the process was down.
I see what you are doing - enter an "unknown" value, which will "reset" the counter, then zero as a reference point, followed by the actual value. I can't do exactly this, because my DB actually contains 9 counters, and typically only one of them will go offline. However, I do know when it goes offline, so the first time I can enter a 'U', and subsequently just zeros. Then when it comes back online it should pick up correctly. On 9/15/2010 7:40 AM, Alex van den Bogaerdt wrote: > ----- Original Message ----- > From: "Philip Peake" <[email protected]> > To: <[email protected]> > Sent: Wednesday, September 15, 2010 4:35 PM > Subject: [rrd-users] Counter resets - how to handle? > > >> I am sampling a counter inside a software process (operations count) >> and plotting the result over time as a graph. >> I am using type COUNTER - since this is what it is. > Good. > >> All works well until for some reason the process has to be restarted. >> At that point rrdtool assumes that the counter overflowed, and shows a >> huge number of operations. > Yep. > >> Any suggestions on how to deal with this? > Solution: see my site, http://www.vandenbogaerdt.nl/rrdtool and click on > RRDtool update tutorial with examples > This will work if you know a reset is about to occur, or just has occured. > > Workaround: set an reasonable maximum for your RRD, so that spikes are > ignored. Use rrdtool tune to change the maximum allowed rate. > > HTH > Alex > > _______________________________________________ > rrd-users mailing list > [email protected] > https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users _______________________________________________ rrd-users mailing list [email protected] https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users
