graphite-dev team mailing list archive

[chrismd <question135957@xxxxxxxxxxxxxxxxxxxxx>]

Question #135957 on Graphite changed:
https://answers.launchpad.net/graphite/+question/135957

    Status: Open => Answered

chrismd proposed the following answer:
That is quite strange. What version of carbon are you running? If its
from trunk do you know what rev? I'm assuming this is a 2-cpu or
hyperthreaded system because exactly 50% cpu usage would be very
strange. Usually when I see that its using up 100% of one logical cpu.
>From your log entries it looks like it is stuck in a loop sorting your
cache queues and emptying them very very slowly.

My best guess at this point is that you may have had a lot of *new*
metrics in the queues because those are rate-limited in terms of how
quickly they can be pulled out (based on MAX_CREATES_PER_MINUTE). I
remember a bug in an older version of carbon where it would behave like
this if only new metrics were in the cache and a MAX_CREATES_PER_MINUTE
was set to be finite and carbon-cache wasn't sleeping between cache
iterations (a sort is done after each iteration so rapidly re-sorting
implies its repeatedly iterating the cache).

If you're unsure what version/rev of carbon, please send me your
carbon/lib/writer.py file and I can check to see if this bug is present
in your version.

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