Date: Wed, 25 Aug 2010 00:01:36 +0900 From: DarkSoul <darksoul@darkbsd.org> To: freebsd-fs@freebsd.org Subject: A way to alter ZFS metadata stored on a drive ? Message-ID: <4C73DED0.7070308@darkbsd.org>
next in thread | raw e-mail | index | archive | help
[-- Attachment #1 --] Hi, I currently am bumping against this bug : http://bugs.opensolaris.org/view_bug.do?bug_id=6782540 Long story short, I had a dead drive on my pool, I tried to replace it only to have the planned spare drive die on me. Which led to this : pool: prana state: DEGRADED scrub: none requested config: NAME STATE READ WRITE CKSUM prana DEGRADED 0 0 0 raidz1 ONLINE 0 0 0 ad16 ONLINE 0 0 0 ad30 ONLINE 0 0 0 ad28 ONLINE 0 0 0 ad18 ONLINE 0 0 0 ad14 ONLINE 0 0 0 raidz1 DEGRADED 0 0 0 ada2 ONLINE 0 0 0 ad4 ONLINE 0 0 0 ad6 ONLINE 0 0 0 ad8 ONLINE 0 0 0 replacing UNAVAIL 0 12.2K 0 insufficient replicas 16391273719868046777 UNAVAIL 0 14.5K 0 was /dev/ada3/old 17286942469715238412 UNAVAIL 0 14.5K 0 was /dev/ada3 raidz1 ONLINE 0 0 0 ada3 ONLINE 0 0 0 ada1 ONLINE 0 0 0 ada0 ONLINE 0 0 0 ad32 ONLINE 0 0 0 ad34 ONLINE 0 0 0 cache ad12 ONLINE 0 0 0 ad37 ONLINE 0 0 0 Here is the zdb dump : prana version=14 name='prana' state=0 txg=3789517 pool_guid=4093564297936254749 hostname='' vdev_tree type='root' id=0 guid=4093564297936254749 children[0] type='raidz' id=0 guid=2955456027802876323 nparity=1 metaslab_array=14 metaslab_shift=36 ashift=9 asize=7501485178880 is_log=0 children[0] type='disk' id=0 guid=14068763453858628066 path='/dev/ad16' whole_disk=0 DTL=706 children[1] type='disk' id=1 guid=3644994118268583474 path='/dev/ad30' whole_disk=0 DTL=4152 children[2] type='disk' id=2 guid=14192589917373989656 path='/dev/ad28' whole_disk=0 DTL=93 children[3] type='disk' id=3 guid=12279545963786751752 path='/dev/ad18' whole_disk=0 DTL=380 children[4] type='disk' id=4 guid=9980201723692421139 path='/dev/ad14' whole_disk=0 DTL=647 children[1] type='raidz' id=1 guid=1078755695237588414 nparity=1 metaslab_array=175 metaslab_shift=36 ashift=9 asize=7501485178880 is_log=0 children[0] type='disk' id=0 guid=12900041001921590764 path='/dev/ada2' whole_disk=0 DTL=4127 children[1] type='disk' id=1 guid=7067019390683734240 path='/dev/ad4' whole_disk=0 DTL=198 children[2] type='disk' id=2 guid=783285185901703545 path='/dev/ad6' whole_disk=0 DTL=197 children[3] type='disk' id=3 guid=9688838378677189049 path='/dev/ad8' whole_disk=0 DTL=196 children[4] type='replacing' id=4 guid=500478404305589015 whole_disk=0 children[0] type='disk' id=0 guid=16391273719868046777 path='/dev/ada3/old' whole_disk=0 not_present=1 DTL=195 children[1] type='disk' id=1 guid=17286942469715238412 path='/dev/ada3' whole_disk=0 not_present=1 DTL=4146 children[2] type='raidz' id=2 guid=4090878701271390479 nparity=1 metaslab_array=164 metaslab_shift=36 ashift=9 asize=7501485178880 is_log=0 children[0] type='disk' id=0 guid=2581994749932752632 path='/dev/ada3' whole_disk=0 DTL=394 children[1] type='disk' id=1 guid=3261576559815570039 path='/dev/ada1' whole_disk=0 DTL=393 children[2] type='disk' id=2 guid=2403801851152895606 path='/dev/ada0' whole_disk=0 DTL=4105 children[3] type='disk' id=3 guid=5795813053837736090 path='/dev/ad32' whole_disk=0 DTL=391 children[4] type='disk' id=4 guid=4547118883658878725 path='/dev/ad34' whole_disk=0 DTL=390 I now can't detach/remove/replace the following drive because "no valid replicas" or because you "can't replace a replacing drive". I gather the problem would be solved if I had a drive with the proper ZDB information and GUID to fool the pool, but I am clueless as to how to generate this information, and it doesn't seem from what the manpage says, that zdb allows you to modify this information. Either that, or find a way to remove these bogus devices, or have the pool "forget" that it must replace a device that does not exist anymore, with a device that does not exist anymore. I would gladly welcome any clue as to how the ZDB metadata work and are stored on the disk, as I can't find written specifications short of probing the ZFS code itself. I can provide any information required, such as the first 512K for each drive (containing the ZDB copies). Thanks in advance for any help. -- Stephane LAPIE, EPITA SRS, Promo 2005 "Even when they have digital readouts, I can't understand them." --MegaTokyo [-- Attachment #2 --] -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAkxz3tYACgkQ24Ql8u6TF2OimACg9g8vHDCE1ndEp7SzB4y3v89O 2bwAoL1zVbjaeH44PLW1mfdVKkHROHDZ =PUe+ -----END PGP SIGNATURE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C73DED0.7070308>