[Openstack-operators] rbd ephemeral storage, very slow deleting...
jeremy at dreamhost.com
Thu Sep 25 04:33:50 UTC 2014
It's normal for them to be slower than other operations, but 18
minutes is *way* too long unless you're running on some very weak
hardware (or using volumes in the terabytes, I guess?). In our
systems, "slow" is considered about 2-5 seconds for a delete on an
80GB volume. That's only "slow" because create operations happen in a
fraction of a second. You might want to strace the rbd client to see
if you're not hitting something bad (max_files is the most likely
limitation being hit during delete operations in my experience).
On Wed, Sep 24, 2014 at 1:29 PM, Abel Lopez <alopgeek at gmail.com> wrote:
> This is expected behavior, unfortunately.
> I spoke to the ceph guys about this last year. When you delete an ‘image’ from a pool, the monitors (IIRC) don’t instantly know where all the segments are across all the OSDs, so it takes a while to find/delete each one.
> On Sep 24, 2014, at 12:45 PM, Jonathan Proulx <jon at jonproulx.com> wrote:
>> Hi All,
>> Just started experimenting with RBD (ceph) back end for ephemeral
>> storage on some of my compute nodes.
>> I have it launching instances just fine, but when I try and delete
>> them libvirt shows the instances are gone, but OpensStack lists them
>> in 'deleting' state and the rbd process on the hypervisor spins madly
>> at about 300% cpu ...
>> ...and now approx 18min later they have finally fully terminated, why so long?
>> OpenStack-operators mailing list
>> OpenStack-operators at lists.openstack.org
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
More information about the OpenStack-operators