graphite-dev team mailing list archive

[Cody Stevens <question170794@xxxxxxxxxxxxxxxxxxxxx>]

Question #170794 on Graphite changed:
https://answers.launchpad.net/graphite/+question/170794

    Status: Needs information => Open

Cody Stevens gave more information on the question:
It is a hardware raid with 6 disks.   I just spoke with the tech in that datacenter.  I thought they were RAID 5 but they are RAID 6.
No BBU's yet read ahead, write back,  256k stripe.

It is xfs file system mounted as follows: 
/dev/sdb1               /scratch                xfs     nobarrier,noatime 0 0

I've been thinking perhaps the retention policy was a contributing
factor so I will go in and pare back the retentions.  It is good to know
that I was heading the right direction,  just had to hear someone else
say it.

However, since this is the initial rollout I know that our metrics are
only going to grow as we find new things we want to measure.  Given that
fact, going forward what would be a recommended configuration to handle
a large amount of metrics?  I guess what would the dream server consist
of?   Obviously,  anything we can build will also be able to be choked
as Kevin mentioned and of course wise retention policies and fast disks
will help.   Just wondering what someone who knows the code would
recommend.

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