graphite-dev team mailing list archive

[Julian Wilde <question187391@xxxxxxxxxxxxxxxxxxxxx>]

New question #187391 on Graphite:
https://answers.launchpad.net/graphite/+question/187391

I've just installed a basic deployment of Graphite 0.9.9 on an EC2 small instance running Ubuntu 11.  Everything looks pretty up to date.

It's running in a steady state; "committedPoints" shows an average reading of about 15K (that's per minute, I guess).  15 min Load avg is 0.23, IO is moderate showing an average of 15% utilization.  Performance seems fine.

But when I perform (what I perceive to be) a basic query, fetching 5 metrics from the last hour (120s resolution) it consistently takes >= 1000ms to fetch the data via HTTP on localhost.  THe size of the data set returned is, obviously, quite small.

Here's what my test URL looks like:
http://statsd/render/?from=-1hour&rawData=true&target=7333.1&target=7333.2&target=7333.3&target=7333.4&target=7333.5

It's running on apache & mod_python.  I see the same results when I use the standalone development server.

This slowness strikes me as odd.  Using whisper-fetch.py to read the relevant .wsp's does so in milliseconds (i.e. it appears to be in the kernel cache).

Is this response time considered normal?  Any suggestions on where I might look to troubleshoot this?



-- 
You received this question notification because you are a member of
graphite-dev, which is an answer contact for Graphite.