[rrd-users] rrd problem

[Canhua]

On Mon, Apr 26, 2010 at 11:59 PM, Simon Hobson <linux at thehobsons.co.uk> wrote:
> Canhua wrote:
>
>> I create a rrd, the info is as follows:
>> _______________________________________
>> filename = "mailrelay1.rrd"
>> rrd_version = "0003"
>> step = 300
>> last_update = 1272293941
>
> ...
>>
>> as you can see the step is 300 seconds, and I update the rrd every minute,
>> but I get NaN when I "fetch" from the rrd
>
> How long did you enter data for ?
> Which RRA did you fetch ?
>
> Each data point (where primary or consolidated) will not be available until
> there has been atr least one update after the period ends. For a primary
> data point, this will be after an integral number of "step" seconds from
> unix epoch has elapsed - so that means every 5 minutes (ie n:00, n:05, n:10,
> ...). For consolidated data points, it means integral multiples of
> (step*pdp_per_row) from unix epoch (eg if pdp_per_row were 3, then that
> would mean on the hour, quarter past, half past, and quarter to). In your
> case, pdp_per_row is 1 so you should get data available when you do an
> update that is on or past each five minute step.
>
> Ie, if you update at 00:04:59 then you have not completed the step and it
> will be NaN. If you update on or after 00:05:00 then it is and data will be
> available for that step.
>
> In each case, you must have entered enough data for the sample period to be
> known. In your case, if you don't enter data for more than 10 minutes (600s)
> then the data becomes unknown.
>
> In your dump, I see unknown_sec = 241 - that suggests you've only just
> started entering data - give it another 5 minutes and it should be OK.


 I enter data every minute by cron, and I fetch the AVERAGE RRA(I have
only one RRA actually). My cron have run many many hours and even
days, and I still 'fetched NaN.