Brian Hechinger
2008-Mar-04 12:41 UTC
[zfs-discuss] What is likely the best way to accomplish this task?
I realize I can''t remove devices from a vdev, which, well, sucks and all, but I''m not going to complain about that. ;) I have 4x500G disks in a RAIDZ. I''d like to repurpose one of them as I''m finding that all that space isn''t really needed and that one disk would serve me much better elsewhere (as the second half of a mirror in a machine going into colo). SYS1 124G 1.21T 29.9K /SYS1 the only other storage in the machine is the small 60G pool made up of the remains of the OS disks, which is not going to be enough space to hold this all while I rebuild the array. Is there any way that I could possibly get one of the disks out of the pool long enough to be used as temp space while I rebuild the array with just 3 disks? I''m thinking either to pull it out of the machine and scribble all over it in another machine so that it no longer has its ZFS bits on it. If this plan works, then it would just be a blank disk (although it will have the same dev id, so I don''t know if zfs will pick it back up anyway). The other plan is to replace it with a file that resides on the 60G partition. This seems more likely to work, but I don''t know how that''d work out with the fact that the file would be a lot smaller than the actual pool. Maybe a thinly provisioned dev would do the trick, as I could then make it look like 500G, but it would only really use what it needed to resync the pool. Is this logic all correct based on how ZFS works on snv_64a? I could also upgrade first if that''s at all recommended, it''s something I''ve been meaning to do anyway. -brian -- "Perl can be fast and elegant as much as J2EE can be fast and elegant. In the hands of a skilled artisan, it can and does happen; it''s just that most of the shit out there is built by people who''d be better suited to making sure that my burger is cooked thoroughly." -- Jonathan Patschke
Tim
2008-Mar-04 14:45 UTC
[zfs-discuss] What is likely the best way to accomplish this task?
On 3/4/08, Brian Hechinger <wonko at 4amlunch.net> wrote:> > I realize I can''t remove devices from a vdev, which, well, sucks and > all, but I''m not going to complain about that. ;) > > I have 4x500G disks in a RAIDZ. I''d like to repurpose one of them as > I''m finding that all that space isn''t really needed and that one disk > would serve me much better elsewhere (as the second half of a mirror > in a machine going into colo). > > SYS1 124G 1.21T 29.9K /SYS1 > > the only other storage in the machine is the small 60G pool made up of > the remains of the OS disks, which is not going to be enough space to > hold this all while I rebuild the array. > > Is there any way that I could possibly get one of the disks out of the > pool long enough to be used as temp space while I rebuild the array with > just 3 disks? > > I''m thinking either to pull it out of the machine and scribble all over > it in another machine so that it no longer has its ZFS bits on it. If > this plan works, then it would just be a blank disk (although it will > have the same dev id, so I don''t know if zfs will pick it back up anyway). > > The other plan is to replace it with a file that resides on the 60G > partition. This seems more likely to work, but I don''t know how that''d > work out with the fact that the file would be a lot smaller than the > actual pool. Maybe a thinly provisioned dev would do the trick, as I > could then make it look like 500G, but it would only really use what > it needed to resync the pool. > > Is this logic all correct based on how ZFS works on snv_64a? > > I could also upgrade first if that''s at all recommended, it''s something > I''ve been meaning to do anyway. > > -brian > -- > "Perl can be fast and elegant as much as J2EE can be fast and elegant. > In the hands of a skilled artisan, it can and does happen; it''s just > that most of the shit out there is built by people who''d be better > suited to making sure that my burger is cooked thoroughly." -- Jonathan > Patschke > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss >I''d suggest going the "pull one disk" route. Stick it into another machine, reformat it with gparted (http://gparted-livecd.tuxfamily.org/), and you''re on your way. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20080304/106e0e9e/attachment.html>
Rob Logan
2008-Mar-04 14:48 UTC
[zfs-discuss] What is likely the best way to accomplish this task?
> have 4x500G disks in a RAIDZ. I''d like to repurpose [...] as the second> half of a mirror in a machine going into colo. rsync or zfs send -R the 128G to the machine going to the colo if you need more space in colo, remove one disk faulting sys1 and add (stripe) it on colo (note: you will need to destroy the pool on colo after copying everything back to attach rather than add the disk going to colo) destroy and remake sys1 as 2+1 and copy it back. removing a vdev is coming, but it will be a hole vdev, ie: remove the 3+1 after you added a 2+1.
Brian Hechinger
2008-Mar-04 15:12 UTC
[zfs-discuss] What is likely the best way to accomplish this task?
On Tue, Mar 04, 2008 at 09:48:05AM -0500, Rob Logan wrote:> > have 4x500G disks in a RAIDZ. I''d like to repurpose [...] as the second > > half of a mirror in a machine going into colo. > > rsync or zfs send -R the 128G to the machine going to the coloYeah, that''s the fallback plan, which I was trying to avoid, but, it might just be the way to go. -brian -- "Perl can be fast and elegant as much as J2EE can be fast and elegant. In the hands of a skilled artisan, it can and does happen; it''s just that most of the shit out there is built by people who''d be better suited to making sure that my burger is cooked thoroughly." -- Jonathan Patschke
MC
2008-Mar-07 02:06 UTC
[zfs-discuss] What is likely the best way to accomplish this task?
> I have 4x500G disks in a RAIDZ. I''d like to repurpose one of them > SYS1 124G 1.21T 29.9K /SYS1This seems to be a simple task because RAID5/Z runs just fine when it is missing one disk. Just format one disk any way that works (take the array offline and do it with format or zpool, or boot into a different os boot cd, etc), copy the 124gb to it from the degraded array, destroy the raidz array, create a new 3 disk array, copy the 124gb back to it, and use the single disk elsewhere. This message posted from opensolaris.org
Brian Hechinger
2008-Mar-07 12:33 UTC
[zfs-discuss] What is likely the best way to accomplish this task?
On Tue, Mar 04, 2008 at 09:48:05AM -0500, Rob Logan wrote:> > have 4x500G disks in a RAIDZ. I''d like to repurpose [...] as the second > > half of a mirror in a machine going into colo. > > rsync or zfs send -R the 128G to the machine going to the coloAnd this is what I ended up doing. So much for the "fancy" solution. ;) -brian -- "Perl can be fast and elegant as much as J2EE can be fast and elegant. In the hands of a skilled artisan, it can and does happen; it''s just that most of the shit out there is built by people who''d be better suited to making sure that my burger is cooked thoroughly." -- Jonathan Patschke