[rrd-users] Can't store data to RRD file

Simon Hobson linux at thehobsons.co.uk
Tue Jan 17 17:15:49 CET 2012

Adriano Monteiro Marques wrote:

>>You've defined an RRD with a step of 1 - that means 1 second per PDP.
>>You've set the heartbeat to 10 in your DSs, so if more than 10
>>seconds between updates the value becomes unknown. Given the
>>timescales in the output, it looks like you are updating at somewhat
>>more than 10 second gaps, therefore your data is always unknown.
>I ran each update command 1 second or so apart, so it shouldn't be 
>missing the data because of that. Did you try the example I 

No, because it is incomplete as I said. You did not provide any 
timing information.

>>Next, you've defined three RAAs, all with consolidation function of
>>AVERAGE, all with 300 steps/row. Thus the RRA with 100 rows contains
>>exactly the same data as the first 100 rows of the other two RRAs,
>>and the RRA with 500 rows contains exactly the same data as the RRA
>>with 1000 rows.
>The RRA shouldn't affect the way data is store anyways. The fact is 
>that is isn't being stored.

So you haven't read, or don't understand, the docs. The **ONLY** data 
you will get out is that defined by an RRA - if you have only put in 
about 10s of data then you will not have any data to extract from an 
RRA with a consolidated step time of 300 seconds (5 minutes) and an 
xff of 0.5 (meaning at least 1/2 the period must have valid data for 
the consolidation to be valid).

>>And of course, what you show us is **NOT** enough to reproduce the
>It is enough. Just run the create command, and run the updates 
>individually, 1 sec or apart from each other and you'll be able to 

And as I said, you provided no timing - so contrary to what you say, 
you did not provide sufficient information.

>>You give no idea at all of timing - "N" tells us nothing
>>about when the updates were run. If you are going to post examples,
>>you need to include that data (ie use actual times rather than "N").
>Although N doesn't tell anything about it, it will use the current 
>localtime. If you run the commands sequentialy, you'll be able to 
>reliably reproduce without having to deal with cumbersome epoch 
>numbers in the command line.

True - but you don't say how far apart to run the commands. I could 
copy/paste and run them all in a fraction of a second. Or I could run 
them hours apart. Whatever I did would be a complete guess.

>>Try Alex's tutorials at 
>>especially "Rates, normalizing and consolidating"
>Thanks for the tips! Do you see anything else I could be missing out?

I think when you've read that tutorial you'll realise why you haven't 
got anything out.
With your numbers, you need to put in several minutes worth of data 
before you'll see any results.

Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

More information about the rrd-users mailing list