On Wed, Jul 25, 2018 at 8:45 AM, Sage Weil <sweil@xxxxxxxxxx> wrote:
> Some of the container-based deployment scenarios bind-mount /var/run (or
> /run these days?) to the host's /var/run. This allows admin
> socket scrapers to run once on the host for all containers. (I'm not sure
> if there are other motivations for doing this.)
>
> In order for that to work, we need to make sure the asok files don't
> collide (e.g., two osd.0's on the host in different clusters/containers
> both trying to use /var/run/ceph/ceph-osd.0.asok).
>
> Ideally, I think, there would be a 'container id' that is unique for each
> container on the host that could be used in the admin_socket config option
> specifying the filename. Does such a thing exist?
>
> Seb also mentioned
>
>> On docker only this works:
>>
>> [root@ceph-nano-lul-faa32aebf00b /]# awk -F '/' '/:memory:\/docker/ \
>> {print $3}' /proc/self/cgroup
>> b80d7e22a089348498710e11be333dd97ed31c000e914ebc03da2b3a76a0aca6
>
> but that seems less than ideal.
Instead of relying on system metadata, perhaps we could use the
instance id? That would require first setting up a MonClient. Maybe
that's acceptable?
--
Patrick Donnelly
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
