From mboxrd@z Thu Jan 1 00:00:00 1970 From: NeilBrown Subject: Re: raid1 issue after disk failure: both disks of the array are still active Date: Fri, 14 Sep 2012 18:13:15 +1000 Message-ID: <20120914181315.79d6d746@notabene.brown> References: <5051AF17.8010501@linuxsystems.it> <20120913103432.GA11764@cthulhu.home.robinhill.me.uk> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=PGP-SHA1; boundary="Sig_/UVnMef381PjCdNP9gIrjAyJ"; protocol="application/pgp-signature" Return-path: In-Reply-To: Sender: linux-raid-owner@vger.kernel.org To: Mikael Abrahamsson Cc: Chris Murphy , Linux RAID List-Id: linux-raid.ids --Sig_/UVnMef381PjCdNP9gIrjAyJ Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable On Fri, 14 Sep 2012 09:16:20 +0200 (CEST) Mikael Abrahamsson wrote: > On Thu, 13 Sep 2012, Chris Murphy wrote: >=20 > > "check" records errors, no action is taken by the md driver to correct= =20 > > it, although the disk firmware itself may try reallocation. So far, tha= t=20 > > appears to not be the case. > > > > "repair" causes the md driver to write correct data (from copy or=20 > > reconstructed from parity), which should force the disk firmware to=20 > > reallocate the affected LBAs from bad physical sectors to good ones. > > > > It seems in this case "repair" is indicated. >=20 > I was under the impression that "check" would check if all data blocks an= d=20 > parity are correct, and record if there is a parity mismatch. This would= =20 > then be corrected by using "repair" at a later time. >=20 > I was also under the impression that if there was a read error on a drive= =20 > during "check", that read error would be corrected using parity because=20 > it's obviously a hard error, not a logical error. Both of your impressions are correct. NeilBrown >=20 > Could you (or someone else) please confirm that my impression is wrong an= d=20 > if there indeed is a hard read error using "check", this will not be=20 > corrected? I would be interested in knowing why this decision was taken t= o=20 > have this behaviour, as I feel that if there is a hard read error, this=20 > should always be corrected using parity. >=20 --Sig_/UVnMef381PjCdNP9gIrjAyJ Content-Type: application/pgp-signature; name=signature.asc Content-Disposition: attachment; filename=signature.asc -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.18 (GNU/Linux) iQIVAwUBUFLnGznsnt1WYoG5AQJ3nw/9GDGWwL5e0N4npABVf5Gub+YxcU6NfipS Ltyq57PNXd02rAj4inmz8Q37433KKTzyz2hKiMebwqrEgksqaQPskv908b1YG50m g0I57WMxGLSq2tmAwhn9D4p0rLrSeWN83aD0OhWg5RUbOFfpyceqDt1vO42w6iqb 3FNvcWZKIN6QelG9XHE9O7wgD6VEw6/wtMORrF2Upt1chI5k0QxI/2xWE0E7FcJO hA53/aYnP3p2auz8QjF1es/lE15+8bylFWnv1wSwnwnXGgw7GGNNG1ZGHSpC4Zqh +QnawXq8L80kWhBNRUfJ3DV/PSwCP0IKgB9pVonXXjvJLIf0opU4bdflwc0VZiLv eGYcr971kzO0JNxmSLPt+U2qEF6rIQ+m7h9I/ylCCq8JqxVJM8NaQJROiFL/EBHj kcZ3me/0Ggp+gNnKfSivxbl7sP93nBHxligPTElnwDqtWooRCty573BZljrMHTYC 6Kb63HqEmj7Ra3DOON361yaGcZ4HKeAczt+EBrgC9vsXudtc1Lxeaz4Pn6waB/up WYmnxePjxcL8Kil+SjNPvK9rkx4+IaG1sMVoztY1RcNKcOw59GG8yIYfBEwGQEiq zztO+cdbPXXr+06UPtDcQedusmGWx3VZKduJwT/N7BDXXMbCyu8zB49DjTvB/cXs ATTaWSMN8O4= =4RAs -----END PGP SIGNATURE----- --Sig_/UVnMef381PjCdNP9gIrjAyJ--