graphite-dev team mailing list archive

[Cody Stevens <question176255@xxxxxxxxxxxxxxxxxxxxx>]

Question #176255 on Graphite changed:
https://answers.launchpad.net/graphite/+question/176255

Cody Stevens gave more information on the question:
Yes... the graph will go for sometimes an hour or more with no holes in
it and then it will be missing some data. I know that if you aren't
sending metrics at the same interval as the retention policy it will
manifest itself in the same way.   At this point it appears to only be
happening to the carbon metrics which are built in.   We have so many it
would be hard to inspect all of them looking for holes so it is hard to
say if it is happening to my other metrics.  I haven't seen it manifest
itself on any of the other graphs I've looked at.   I ran into this
issue when I was I/O bound by disk before but the cache was filling
completely in that case.  It was also easier to spot because all the
graphs would have holes in them for the same time frame while the cache
was full.  This is something new ( to me anyway).  I don't think it is
cpu bound the load on the box is pretty low and I have plenty of memory.
I think it is interesting that it is only happening to the carbon
metrics which I would expect to be the most stable since they don't even
have to cross the network to get to the cache.

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