Move WAL/DB to SSD for existing OSD?
Nautilus - Bluestore OSD's created with everything on disk. Now I have some spare SSD's - can I move the location of the existing WAL and/or DB to SSD partitions without recreating the OSD? I suspect not - saw emails from 2018, in the negative :( Failing that - is it difficult to add lvmcache to a osd? -- Lindsay
Hi Lindays, Am 29.06.20 um 15:37 schrieb Lindsay Mathieson:
Nautilus - Bluestore OSD's created with everything on disk. Now I have some spare SSD's - can I move the location of the existing WAL and/or DB to SSD partitions without recreating the OSD?
I suspect not - saw emails from 2018, in the negative :(
Yes you can. You need to use: ceph-bluestore-tool --path /var/lib/ceph/osd/ceph-${OSD} bluefs-bdev-new-db --dev-target /dev/bluesfs_db/db-osd${OSD} and ceph-bluestore-tool --path dev/osd1/ --devs-source dev/osd1/block --dev-target dev/osd1/block.db bluefs-bdev-migrate those are not tested and just what i remembered but it should point you in the richt direction. greets, Stefan
Failing that - is it difficult to add lvmcache to a osd?
On 29/06/2020 11:44 pm, Stefan Priebe - Profihost AG wrote:
You need to use: ceph-bluestore-tool --path /var/lib/ceph/osd/ceph-${OSD} bluefs-bdev-new-db --dev-target /dev/bluesfs_db/db-osd${OSD}
and
ceph-bluestore-tool --path dev/osd1/ --devs-source dev/osd1/block --dev-target dev/osd1/block.db bluefs-bdev-migrate
those are not tested and just what i remembered but it should point you in the richt direction.
greets, Stefan
Thanks Stephan, thats great. Best I test with a single OSD first :) I presume it needs to be down for this. Going to have to wait a bit while the cluster rebalances after a pg reduction though. Might start a separate thread on lvmcache versus WAL/DB on SSD for VM usage, if that hasn't been done to death... -- Lindsay
Don't forget to set the correct LV tags for the new db device as mentioned in [1] and [2]. [1] https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/6OHVTNXH5SLI... [2] https://tracker.ceph.com/issues/42928 Zitat von Lindsay Mathieson <lindsay.mathieson@gmail.com>:
On 29/06/2020 11:44 pm, Stefan Priebe - Profihost AG wrote:
You need to use: ceph-bluestore-tool --path /var/lib/ceph/osd/ceph-${OSD} bluefs-bdev-new-db --dev-target /dev/bluesfs_db/db-osd${OSD}
and
ceph-bluestore-tool --path dev/osd1/ --devs-source dev/osd1/block --dev-target dev/osd1/block.db bluefs-bdev-migrate
those are not tested and just what i remembered but it should point you in the richt direction.
greets, Stefan
Thanks Stephan, thats great. Best I test with a single OSD first :) I presume it needs to be down for this. Going to have to wait a bit while the cluster rebalances after a pg reduction though.
Might start a separate thread on lvmcache versus WAL/DB on SSD for VM usage, if that hasn't been done to death...
-- Lindsay _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
On 30/06/2020 8:17 pm, Eugen Block wrote:
Don't forget to set the correct LV tags for the new db device as mentioned in [1] and [2].
[1] https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/6OHVTNXH5SLI...
Thanks and *ouch* - thats a lot of manual tweaking with things I don't really understand if stuff goes wrong and is still in testing/experimental? I might just setting for slowly recreating the OSD's - seems safer. Its not like I'm in a hurry or will be traveling anywhere soon :) Thanks! -- Lindsay
participants (3)
-
Eugen Block
-
Lindsay Mathieson
-
Stefan Priebe - Profihost AG