← Back to team overview

graphite-dev team mailing list archive

Re: [Question #158830]: MAX_CREATES_PER_MINUTE discarding metrics

 

Question #158830 on Graphite changed:
https://answers.launchpad.net/graphite/+question/158830

    Status: Needs information => Solved

ziggy confirmed that the question is solved:
Sorry I missed answering some of your questions.  Those graphs are
unremarkable...avgUpdateTime is very low and updateOperations is
similarly uninteresting (I don't have it in front of me but looked at
them earlier).  These are all new unique metrics I'm testing with, and I
have started (several times) with a clean slate.

2) Yes we are using Memcached.

3) I need about 30,000 updates per second to one machine, but it is
bursty traffic so it doesn't translate to 600k per minute.  It actually
comes as 50k in about 2 seconds with a tough upper bound on the time.
Currently, pyped takes this data in about 2 seconds on old hardware so
we're hoping to do at least that good.

Hope this helps.  I still am baffled by the huge performance increase on
my staging box when I raised MAX_CREATES because when it hits that it
just pops the queue which you wouldn't think would be a slow operation.
I'm going to read over the source code some more tomorrow.  Thanks for
your time Nicholas!

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