G''day, My OpenSolaris (b134) box is low on space and has a ZFS mirror for root : uname -a SunOS wattage 5.11 snv_134 i86pc i386 i86pc rpool 696G 639G 56.7G 91% 1.09x ONLINE - It''s currently a pair of 750GB drives. In my bag I have a pair of brand spanking new 2TB seagates that I plan to replace the 750''s with. Can anyone here confirm that the following will (should!) work : take one drive offline shutdown replace physical drive & restart add it to the mirror, wait for the mirror to resilver installboot on the new drive boot check everything take other drive offline shutdown replace physical drive & restart add to the mirror, wait for resilver installboot on new drive run zpool list/df etc and see LOTS more disk space in rpool! Anything I may have missed? -- This message posted from opensolaris.org
On 09/23/10 05:00 PM, Carl Brewer wrote:> G''day, > My OpenSolaris (b134) box is low on space and has a ZFS mirror for root : > > uname -a > SunOS wattage 5.11 snv_134 i86pc i386 i86pc > > > rpool 696G 639G 56.7G 91% 1.09x ONLINE - > > It''s currently a pair of 750GB drives. In my bag I have a pair of brand spanking new 2TB seagates that I plan to replace the 750''s with. > > Can anyone here confirm that the following will (should!) work : > > take one drive offline > shutdown > replace physical drive& restart > add it to the mirror, wait for the mirror to resilver > installboot on the new drive > boot > check everything > take other drive offline > shutdown > replace physical drive& restart > add to the mirror, wait for resilver > installboot on new drive > > run zpool list/df etc and see LOTS more disk space in rpool! > > Anything I may have missed? >Make a cup of tea while the mirrors resilver! Your steps look to be complete. -- Ian.
Ok, that doesn''t seem to have worked so well ... I took one of the drives offline, rebooted and it just hangs at the splash screen after prompting for which BE to boot into. It gets to hostname: blah and just sits there. Um ... I read some doco that says : The boot process can be slow if the boot archive is updated or a dump device has changed. Be patient. (from http://www.solarisinternals.com/wiki/index.php/ZFS_Troubleshooting_Guide#Disk_Replacement_Example) DO I need to just wait and it''ll come up eventually or do I need to do something more constructive? There''s no drive activity after an initial flurry for ~1-2 mins or so. It''s just hanging. I kinda need this box up, nothing on it is irreplaceable but it''ll be a major PITA if I lose it. Can the root pool be mounted in some form on another box and the data from it be retreived? It''s running a more current zpool version than 200906 (whatever''s current in b134) - maybe if I crank up an openIllumos or whatever it''s called and see if that can mount the drives somehow? -- This message posted from opensolaris.org
it is responding to pings, btw, so *something''s* running. Not ssh though .... -- This message posted from opensolaris.org
swapping the boot order in the PC''s BIOS doesn''t help.... -- This message posted from opensolaris.org
Casper.Dik at Sun.COM
2010-Sep-23 13:06 UTC
[zfs-discuss] Growing a root ZFS mirror on b134?
>Ok, that doesn''t seem to have worked so well ... > >I took one of the drives offline, rebooted and it just hangs at the >splash screen after prompting for which BE to boot into. >It gets to >hostname: blah >and just sits there.When you say "offline", did you: - remove the drive physically? - or did you zfs detach it? - or both? In order to remove half of the mirror I suggest that you: split the mirror (if your ZFS is recent enough; seems to be supported since 131) [ make sure you remove /etc/zfs/zpool.cache from the split half of the mirror. ] or detach only then remove the disk. Depending on the hardware it may try to find the missing disk and this may take some time. You can boot with the debugger and/or -v to find out was is going on. Casper
On 23/09/2010 11:06 PM, Casper.Dik at Sun.COM wrote:> >> Ok, that doesn''t seem to have worked so well ... >> >> I took one of the drives offline, rebooted and it just hangs at the >> splash screen after prompting for which BE to boot into. >> It gets to >> hostname: blah >> and just sits there. > > > When you say "offline", did you: > > - remove the drive physically? > - or did you zfs detach it? > - or both?zpool offline rpool <drive> It''s plugged back in now (I''m trying all sorts of things!)> In order to remove half of the mirror I suggest that you: > > > split the mirror (if your ZFS is recent enough; seems to be > supported since 131) > [ make sure you remove /etc/zfs/zpool.cache from the > split half of the mirror. ] > or > detach > > > only then remove the disk. > > Depending on the hardware it may try to find the missing disk and this > may take some time."some time" being a minute, an hour? How long should I wait before giving in and trying something else?> > You can boot with the debugger and/or -v to find out was is going on.How is this done on a PC? On SPARC I''d just have said ''boot -s" or whatever the arguments are for that these days, but x86 PC''s? Thanks Casper (I remember you from the 1990''s and your early Solaris 2.x FAQ!) -- This message posted from opensolaris.org
> On 23/09/2010 11:06 PM, Casper.Dik at Sun.COM wrote: > > > >> Ok, that doesn''t seem to have worked so well ... > >> > >> I took one of the drives offline, rebooted and it > just hangs at the > >> splash screen after prompting for which BE to boot > into. > >> It gets to > >> hostname: blah > >> and just sits there. > > > > > > When you say "offline", did you: > > > > - remove the drive physically? > > - or did you zfs detach it? > > - or both? > > zpool offline rpool <drive> > > It''s plugged back in now (I''m trying all sorts of > things!) > > > > In order to remove half of the mirror I suggest > that you: > > > > > > split the mirror (if your ZFS is recent enough; > seems to be > > supported since 131) > > [ make sure you remove /etc/zfs/zpool.cache > from the > > split half of the mirror. ] > > or > > detach > > > > > > only then remove the disk. > > > > Depending on the hardware it may try to find the > missing disk and this > > may take some time. > > "some time" being a minute, an hour? How long should > I wait before giving in and trying something else? > > > > > You can boot with the debugger and/or -v to find > out was is going on. > > How is this done on a PC? On SPARC I''d just have > said ''boot -s" or whatever the arguments are for that > these days, but x86 PC''s? > > Thanks Casper (I remember you from the 1990''s and > your early Solaris 2.x FAQ!)Hi-- I would re-connect the original disk and re-attach it to your rpool. After it resilvers, re-apply the boot blocks. Do you really need such a large root pool (2 TBs)? You might consider creating a separate data pool with the 2 TB disks and move some of your non-OS root pool data over to the data pool to reduce your root pool space consumption. If you still want to increase your root pool by using the 2 TB disks, start over with the replacement process. In general, you would replace a smaller root pool disk by using one of the following options: 1. Attach the larger disk, let it resilver, apply the bootblocks, confirm that you can boot from it. Then, detach the smaller disk. 2. Replace the smaller disk with the larger disk, one disk at a time, by using the zpool replace command. This process is a bit riskier if anything is wrong with the replacement disk. Do you have enough slots to connect all 4 disks at once (the two smaller ones and the two larger ones)? If so, I would recommend a combination of the above options because your root pool is already mirrored. Something like this: disk1 = 750 GB disk2 = 750 GB disk3 = 2 TB disk4 = 2 TB # zpool replace rpool disk2 disk3 /* Use zpool status to check resilvering status /* Apply bootblocks to disk3 /* Test booting from disk3 /* If disk3 boot succeeds, continue: # zpool attach rpool disk3 disk4 /* Use zpool status to check resilvering status /* Apply bootblocks to disk4 /* Test booting from disk4 # zpool detach rpool disk1 You might review some of the steps here: http://www.solarisinternals.com/wiki/index.php/ZFS_Troubleshooting_Guide Replacing/Relabeling the Root Pool Disk Thanks, Cindy -- This message posted from opensolaris.org
Thank you very much to Casper for his help. I now have a rpool with 2 x 2TB HDD''s and it''s all good : # zpool list rpool NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT rpool 1.82T 640G 1.19T 34% 1.09x ONLINE - It gets quite unhappy if you don''t use detach (offline isn''t so useful!) and booting with -m milestone=none is a lifesaver! -- This message posted from opensolaris.org