Not really the place for it I know, but this fora seems a little more well read than the PNP4Nagios example and I'm fairly confident that someone will have some experience of sizing rrds. I've set some RTT checks up on nagios at 15s intervals and the PNP4Nagios plugin is doing a sterling job in graphing the results (i.e. I get some graphs showing data at a granularity of 15 sec intervals), however I'm not convinced the data I am seeing is correct (the reason being that yesterday I saw an alert for a link running at 44ms, but the graph was depicitng the data at something more around the order of 24ms or something)
I think this is possibly down to the configuration file 'RRA.CFG' which specifies the entries built into the rrd file on creation, but can't be sure as I don't understand fully what the configuration is specifying - for example,
# 2880 entries with 1 minute step = 48 houres
I know that 0.25 is the 'fudge factor', i.e. we can take 25% of the entries being missing as an acceptable loss, and obv 2280 is 4 hrs worth of entries if they are 1 min apart... but what if I'm polling at 15 secs? What significance is the "1"? As the plugin was by default configured to use 1 minute checks, the describing text of "1 minute step" strikes me as a bit of a double edged sword...
As far as I can see I have 2 options - either
I) The 1 indicates actual 1 minute time intervals and rrdtool takes into account one entry from the 4 logged every minute, and therefore I need to change this to 0.25 to get it to take into account readings for every 15 secs and boost total entries to 9120 retain the 4 hrs worth of data, as currently I am missing 3/4 of the data
II) The 1 indicates 1 step (or 1 entry, i.e. it's reading every entry), so I only need to change the final total entries to 9120 retain the 4 hrs worth...
So I get (I think) that I'm going to have to rebuild the rrds and that I need to at least change the last figure to 9120, but what of the '1'... does it stay 1, or change to 0.25? What say you?
Any help much appreciated