backfilling or historical reporting

Asked by anthony acquanita

I have as a schema
10 seconds for 5 days and
15min for 1 year

I'm trying to backfill the lost data from over 5 days ago
Sun Feb 12 06:41:30 2012 None
Sun Feb 12 06:41:40 2012 None
Sun Feb 12 06:41:50 2012 None
Sun Feb 12 06:42:00 2012 None
Sun Feb 12 06:42:10 2012 None
Sun Feb 12 06:42:20 2012 None
Sun Feb 12 06:42:30 2012 None
Sun Feb 12 06:42:40 2012 None
Sun Feb 12 06:42:50 2012 None
Sun Feb 12 06:43:00 2012 None
Sun Feb 12 06:43:10 2012 None

Thu Feb 16 12:37:40 2012 611.000000
Thu Feb 16 12:37:50 2012 770.000000
Thu Feb 16 12:38:00 2012 686.000000
Thu Feb 16 12:38:10 2012 685.000000

This is what I'm trying to send
ads01_b.avgTt 639 1329046972 ( Sun Feb 12 2012 06:42:52)
Do I have to make the epoc time match the dates in the whisper file?

Is this how it's supposed to work, am I doing something wrong?

Question information

Language:
English Edit question
Status:
Solved
For:
Graphite Edit question
Assignee:
No assignee Edit question
Solved by:
anthony acquanita
Solved:
Last query:
Last reply:
Revision history for this message
Michael Leinartas (mleinartas) said :
#1

Yeah, thats how it should work. Whisper will round back to the nearest multiple of the precision (10). Usually the only thing to note is that you need to roll-up older metrics yourself if you're beyond the first schema definition (in your case, add or average the 10 second measurements in 15 minute groups once youre backfilling data older than 5 days).

Are you writing to carbon or directly to the whisper files using whisper-update.py? If you're going through carbon and the setup happens to be busy, you could have metrics that are sitting in the cache for a few minutes. If you check the metrics through the web interface, it'll grab from the cache as well as disk and may be a better verifier.

Revision history for this message
anthony acquanita (acqant) said :
#2

I'm writing to carbon.

It seems to be showing up today. I added logging for updates and they are registering.

Thanks.