[rrd-users] Why is this happening?

[Simon Hobson]

Mike_K <mike at kalist.ca> wrote:

A message without a question !

I can **GUESS** from the subject line that **SOMETHING** isn't happening as you would expect. You haven't said that that something is so it's all guesswork.

I would GUESS that you aren't expecting the gaps in the graph, is that correct ?
Why, well we can't say because you haven't given much of a clue what data you fed in. You've provided one update statement using "N" as the timestamp which is meaningless. So it boils down to "I fed in some data I'm not giving you a clue about, and didn't get out what I expect".

To avoid future responses along these lines, you would do well to read and head the advice here :
http://www.catb.org/~esr/faqs/smart-questions.html


Now, assuming that my guesswork is correct, the usual reason for missing data is because you've not supplied data often enough. You have created a database with a step size of 150, and a maximum interval between updates of 300s. If you are doing something like having cron run your script every 5 minutes, then you will get gaps in the data.

Lets say you supply data at 70s and 371s (and with data having been valid before that). At time 70 you have nearly half filled one PDP (primary data point) covering time 0 to 150. At time 370, the data becomes unknown so when the next update arrives, that PDP will be set to unknown (not enough data to complete it), as will the one from 150 to 300 (no data at all). At time 371, you supply data, so this sets a base value against which the rest of the 300 to 450 PDP can be calculated - IFF you provide another update within the next 300s.

For more information, I suggest Alex's excellent tutorials at http://rrdtool.vandenbogaerdt.nl, particularly the one on "Rates, normalizing and consolidating".