Hi Wido, > On 26 Sep 2014, at 23:14, Wido den Hollander <wido at 42on.com> wrote: > > On 26-09-14 17:16, Dan Van Der Ster wrote: >> Hi, >> Apologies for this trivial question, but what is the correct procedure to replace a failed OSD that uses a shared journal device? >> >> Suppose you have 5 spinning disks (sde,sdf,sdg,sdh,sdi) and these each have a journal partition on sda (sda1-5). Now sde fails and is replaced with a new drive. Where do you put the journal for the new sde? It _should_ be sda1 of course, but I didn?t yet work out the correct way to reuse that. >> >> Using ceph-disk (in dumpling), I found that ceph-disk prepare /dev/sde /dev/sda will create a 6th partition on sda. If I rm partition 1 before running ceph-disk, it seems to re-use partition 1 but the udev triggers (probably partx) don?t quite like this and the osd is never activated. >> >> I?m just curious, for such a routine operation, what are most admins doing in this case? >> > > I simply would do: > > $ ceph-deploy osd create mycephnode1:sde:sda1 > > That should work just fine. I forgot to mention I?d tried this as well. Unfortunately this (well, ceph-disk prepare /dev/sde /dev/sda1) creates a journal link to /dev/sda1 instead of the by-partuuid link. (So this isn?t 100% reboot safe). BTW, when re-using a journal do we need to dd zeroes to anywhere in sda1? Cheers, Dan
 |