Steve; Thanks for your reply!
Just to clarify, a metric like CPU Busy, Number of JOBS etc would be better off as a COUNTER? Deepest Regards Steven Sim On Tue, Jun 10, 2014 at 9:09 AM, Steve Shipway <[email protected]> wrote: > Since your first RRA has a count of 1, then 1cdp = 1pdp = 15min. There > are 96 rows, and so (as you say) this RRA has a length of 96x15min = 1 day. > > > > Your second RRA has a count of 4, so 1cdp = 4pdp = 1hour. There are 24 > rows, so it also has a length of 1 day (in which case it is likely not > necessary). > > > > Probably, you want to make these RRA much longer; 2880 rows in the first > RRA will be about 30 days long. > > > > If your metrics are all coming in the same file, for the same point in > time, and all being pushed into the RRD at the same time, then it makes > sense to have a single RRD to hold them as in your example. You would > usually use a separate RRD if the data came separately, potentially for > different times. Then separate RRD would make sense as you may get one > sample but not another, or they were sampled at differing times. > > > > However, in your example, your TapeRate and DASDRate variables have been > given a valid range of 0 – 100. This makes sense if they are a percentage, > but not if they are data rates which could be larger than 100B/s. You may > want to make the upper bound on these DS higher or undefined. Also, if you > can extract a data counter rather than a rate, you will get more accurate > results overall by using this with a DS type of ‘counter’ and allowing > RRDTool to calculate the rate. > > > > Steve > > > > > > *Steve Shipway* > > [email protected] > > > > *From:* [email protected] [mailto: > [email protected]] *On Behalf Of *Steven Sim > *Sent:* Tuesday, 10 June 2014 12:57 p.m. > *To:* Steve Shipway > *Cc:* [email protected] > *Subject:* Re: [rrd-users] If input is already in text format and I craft > a perl script to parse the text format and update rrd database, what should > the step and heartbeat be? > > > > Steve; > > > > Firstly, thanks deeply for replying! > > > > I do have the sample times in my file and this can be easily converted to > epoch time using perl for each data point. > > > > Two more questions; > > > > The data measures mainframe CPU Busy and other metrics, much like Solaris > prstat or Linux top command with each data point 15 minutes apart. > > > > I've created a database like so ... > > > > rrdtool create ${RRDBfile} \ > > --step 900 \ > > --start ${STARTIME} \ > > DS:CPUBusy:GAUGE:1800:0:100 \ > > RRA:AVERAGE:0.5:1:96 \ > > RRA:AVERAGE:0.5:4:24 \ > > > > The ${STARTIME} variable is calculated as the first sample time minus 900 > seconds. > > > > The first RRA archive is 1 PDP with 96 CDP or 24 hours. Is my > understanding correct? > > > > (Since 1 PDP = 15 minutes, 96 CDP will be 15 minutes * 96 or 24 hours) > > > > So if I wish to store for a month, then i do > > > > RRA:AVERAGE:0.5:1:2880 ?? > > > > As for the other metrics, like Tape Rate, DASD Rate, can I storage them in > a single RRD database or would you recommend I create a database each for > each metric? > > > > For example, my understanding for a multi metric database creation would > be .. > > > > rrdtool create ${RRDBfile} \ > > --step 900 \ > > --start ${STARTIME} \ > > DS:CPUBusy:GAUGE:1800:0:100 \ > > DS:DASDRate:GAUGE:1800:0:100 \ > > DS:TAPERate:GAUGE:1800:0:100 \ > > RRA:AVERAGE:0.5:1:96 \ > > RRA:AVERAGE:0.5:4:24 \ > > > > Would the above understanding be correct? > > > > Deepest Regards > > Steven Sim > > On Tue, Jun 10, 2014 at 6:22 AM, Steve Shipway <[email protected]> > wrote: > > If the readings are taken at 15min intervals, then your step should be > 15min (900s), because this is the step between samples. > > > > Generally speaking, unless you have a special case, you should set your > heartbeat to be twice your step – in this case, 1800s > > > > When you update your RRDtool database, you will give RRDTool a point in > time and the data to store at that point in time. DO NOT use ‘N’ (‘now’) > for your point in time as you should be specifying the point in time when > the data were sampled, not when you store them. The step and heartbeat > apply to the data point in time, not when you happen to store them. > > > > Obviously, you need to have the sample times in your text file. If you do > not have them, you may be able to deduce the time by knowing that they are > 15min apart and extrapolating? > > > > Steve > > > > *Steve Shipway* > > [email protected] > > > > *From:* [email protected] > [mailto:[email protected]] *On > Behalf Of *Steven Sim > *Sent:* Tuesday, 10 June 2014 3:12 a.m. > *To:* [email protected] > *Subject:* [rrd-users] If input is already in text format and I craft a > perl script to parse the text format and update rrd database, what should > the step and heartbeat be? > > > > Hello; > > > > my sensor tech staff has collated the readings onto several text files, > each reading 15 minutes apart. > > > > I have crafted a Perl script to parse the text file and execute rrdtool > update for each reading in the text file. > > > > In this case, what should be my step and heartbeat? > > > > The actual readings are taken 15 minutes apart, but the actual feeding > into the rrdtool database are actually loops through the perl script, each > update at most a second or less apart > > > > Do i still retain step=15 minutes and heartbeat=15 minutes x 2 ? > > > > Deepest Regards > Steven Sim > > > > >
_______________________________________________ rrd-users mailing list [email protected] https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users
