[openstack-dev] [Infra] Question about disk partition
oomichi at mxs.nes.nec.co.jp
Mon Jan 27 01:54:24 UTC 2014
> -----Original Message-----
> From: Jeremy Stanley [mailto:fungi at yuggoth.org]
> Sent: Saturday, January 25, 2014 7:52 PM
> To: openstack-dev at lists.openstack.org
> Subject: Re: [openstack-dev] [Infra] Question about disk partition
> On 2014-01-23 13:55:21 -0800 (-0800), Andrew Woodward wrote:
> > I would think that the intent of the tests would be to use the /
> > opt partition that was attached to the instance. I would think
> > that we should be able to configure the resources to use /opt
> > instead.
> It sounds to me like another manifestation of
> https://launchpad.net/bugs/1268732 which we're addressing through
> several patches which hit much of what you suggest:
> * https://review.openstack.org/68782 (don't copy logs to /home)
> * https://review.openstack.org/68990 (git clones in /opt)
> * https://review.openstack.org/69075 (make query logging optional)
Thank you for doing it, now root partition usage has been reduced(almost 80% -> 10%):
Filesystem Size Used Avail Use% Mounted on
/dev/xvda1 40G 4.3G 34G 12% /
udev 3.9G 12K 3.9G 1% /dev
tmpfs 1.6G 312K 1.6G 1% /run
none 5.0M 0 5.0M 0% /run/lock
none 3.9G 4.0K 3.9G 1% /run/shm
/dev/xvde2 72G 9.8G 58G 15% /opt
cgroup 3.9G 0 3.9G 0% /sys/fs/cgroup
/dev/loop0 6.0G 68M 6.0G 2% /opt/stack/data/swift/drives/sdb1
> > or hack /var to be inside /opt.
> Not particularly trivial without a reboot, since getting file
> descriptors closed out on /var at a multi-user runevel is, well, not
> something I've ever had much luck at.
> > Either way, growing the root volume would likely be the most
> > painful route we could take.
> Not just painful, but mostly impossible. Our providers give us a
> tiny block device formatted and mounted as the root filesystem, plus
> a large ephemeral block device which we partition into /opt and swap
> when building a node from a snapshot.
I got it, thanks for the explanation also.
More information about the OpenStack-dev