graphite-dev team mailing list archive

[Cody Stevens <question217749@xxxxxxxxxxxxxxxxxxxxx>]

Question #217749 on Graphite changed:
https://answers.launchpad.net/graphite/+question/217749

    Status: Answered => Open

Cody Stevens is still having a problem:
Restarting the cache on that server did make a difference in the iowait
which is consistently below ~50% .  This is the target we were trying to
hit when tweaking the server and MAX_UPDATES_PER_SECOND based on other
answers.  I'm wondering if the writes could be getting larger over time
and causing the iowait to grow.  That is to say 1 write becomes
significantly more expensive.  We based our MAX_UPDATES_PER_SECOND off
of the average update times reported by carbon.  Of course, those update
times were from 0.9.8 and we are now on 0.9.10.  Also if the update
times were significantly lower during say the first day of the cache
running vs. a cache that has been running for a month perhaps that would
indicate the MAX_UPDATES_PER_SECOND should be set slightly higher.  Just
throwing some theories out there.   Thanks for your help.

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