graphite-dev team mailing list archive

Thread
Date

Re: [Question #133636]: queue size - writes don't keep up with volume

To: graphite-dev@xxxxxxxxxxxxxxxxxxx
From: bhardy <question133636@xxxxxxxxxxxxxxxxxxxxx>
Date: Mon, 15 Nov 2010 06:27:10 -0000
Reply-to: question133636@xxxxxxxxxxxxxxxxxxxxx
Sender: bounces@xxxxxxxxxxxxx

Question #133636 on Graphite changed:
https://answers.launchpad.net/graphite/+question/133636

bhardy posted a new comment:
I thank you very much for the thorough explanation.

Our two systems are Centos x86_64 with 8 cpus at 2327, 31g of memory, and as I mentioned SAN mounts.  I did take the default values in carbon.conf as "recommended settings".  I imagined our systems could handle the volume as I noticed in a former answer on lauchpad you stated you had a system handling 250k metrics/min.
 
After initial population we do not frequently receive a bunch of new metrics. For the first 3-4days of turning things up it will need to build ALOT of new metrics.  After that it should be primarily updates.  Maybe we can just turn things up slowly.  This issue has been killing us so I will apply the adjustments to carbon.conf and test.  I will also see how feasible it is to install the latest version.  Thanks again.

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