On Tue, 4 Dec 2012 10:59:53 -0300 Salatiel Filho <salatiel.filho@xxxxxxxxx> wrote: > On Mon, Nov 26, 2012 at 9:13 PM, NeilBrown <neilb@xxxxxxx> wrote: > > On Sun, 25 Nov 2012 10:30:10 -0300 Salatiel Filho <salatiel.filho@xxxxxxxxx> > > wrote: > > > >> On Tue, Nov 20, 2012 at 6:25 PM, NeilBrown <neilb@xxxxxxx> wrote: > >> > > >> > On Tue, 20 Nov 2012 14:00:35 -0300 Salatiel Filho <salatiel.filho@xxxxxxxxx> > >> > wrote: > >> > > >> > > []'s > >> > > Salatiel > >> > > > >> > > > >> > > On Tue, Nov 20, 2012 at 8:02 AM, Salatiel Filho > >> > > <salatiel.filho@xxxxxxxxx> wrote: > >> > > > Thanks Neil ! > >> > > > []'s > >> > > > Salatiel > >> > > > > >> > > > > >> > > > On Mon, Nov 19, 2012 at 10:26 PM, NeilBrown <neilb@xxxxxxx> wrote: > >> > > >> On Sun, 18 Nov 2012 10:42:22 -0300 Salatiel Filho <salatiel.filho@xxxxxxxxx> > >> > > >> wrote: > >> > > >> > >> > > >>> Hi guys, i will be building a few arrays very soon and i'd like to > >> > > >>> test the badblock feature and the replaceable feature, but i haven`t > >> > > >>> found where i can download the source of mdadm 1.3. Where can i get it > >> > > >>> ? > >> > > >>> > >> > > >> You can't. 3.3 doesn't exist yet. > >> > > >> > >> > > >> However > >> > > >> git clone git://neil.brown.name/mdadm > >> > > >> cd mdadm > >> > > >> make > >> > > >> > >> > > >> will get you code that supports bad blocks and replaceable devices. > >> > > >> > >> > > >> Alternately you can get a tar file at > >> > > >> > >> > > >> http://git.neil.brown.name/git?p=mdadm.git;a=snapshot;h=master > >> > > >> > >> > > >> NeilBrown > >> > > > >> > > Forgot to send a copy to the list: > >> > > > >> > > > >> > > Now the question, how do i enable badblock list ? > >> > > # mdadm -A /dev/md1 /dev/sd[bcd]2 --update=bbl ? > >> > > >> > Yes, I think that is right - does it work? > >> > > >> > > > >> > > I actually created also a new array > >> > > # mdadm -C -n4 -l5 -e 1.2 /dev/md2 /dev/sd[bcde]3 , > >> > > but how do i verify if badblock support was enabled ? > >> > > >> > mdadm -E /dev/sdb3 > >> > > >> > should show > >> > > >> > Bad Block Log : 512 entries available at offset xxx sectors > >> > > >> > NeilBrown > >> > > >> > > -- > >> > > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > >> > > the body of a message to majordomo@xxxxxxxxxxxxxxx > >> > > More majordomo info at http://vger.kernel.org/majordomo-info.html > >> > > >> > >> > >> Well, about the tests: > >> I have put some data on a degraded raid 5 with 3 disks where one of > >> them has random read errors. After that i added a new disk and the > >> raid started to resync. > >> After 62.5% i started getting read errors on /dev/sdd2 and the array > >> stopped to resync. > > > > Any useful messages from the kernel log? > > > >> > >> 1) Shouldn`t badblocks be ignored to allow the recover of most data? > >> I can see in mdadm -E /dev/sdd2 Bad Block Log : 512 entries > >> available at offset 262128 sectors - bad blocks present. > > > > That's the theory. And bad blocks found should be recorded on both sdd2 and > > the new device. > > > >> > >> 2) After stop the array and try to reassembled with mdadm -A > >> /dev/md1 /dev/sd[bcde]2 --force i got the following message: > >> mdadm -A /dev/md1 /dev/sd[bcde]2 --force > >> mdadm: forcing event count in /dev/sdb2(0) from 87705 upto 87713 > >> mdadm: forcing event count in /dev/sdc2(1) from 87705 upto 87713 > >> mdadm: /dev/md1 has been started with 3 drives (out of 4). > > > > That doesn't look good. Again, complete kernel messages might help. > > > >> > >> and i noticed that sde2 was not added to the array. > >> > >> # mdadm -D /dev/md1 > >> /dev/md1: > >> Version : 1.2 > >> Creation Time : Tue Nov 20 08:43:35 2012 > >> Raid Level : raid5 > >> Array Size : 3918320640 (3736.80 GiB 4012.36 GB) > >> Used Dev Size : 1306106880 (1245.60 GiB 1337.45 GB) > >> Raid Devices : 4 > >> Total Devices : 3 > >> Persistence : Superblock is persistent > >> > >> Intent Bitmap : Internal > >> > >> Update Time : Sun Nov 25 10:04:06 2012 > >> State : active, degraded > >> Active Devices : 3 > >> Working Devices : 3 > >> Failed Devices : 0 > >> Spare Devices : 0 > >> > >> Layout : left-symmetric > >> Chunk Size : 512K > >> > >> Name : zotac:1 (local to host zotac) > >> UUID : 93e55a39:94681b9d:4c0a1a15:cad51a71 > >> Events : 87713 > >> > >> Number Major Minor RaidDevice State > >> 0 8 18 0 active sync /dev/sdb2 > >> 1 8 34 1 active sync /dev/sdc2 > >> 2 8 50 2 active sync /dev/sdd2 > >> 6 0 0 6 removed > >> > >> > >> After add it manually , mdadm -a /dev/md1 /dev/sde2 > >> the rebuild started from the beginning, ignoring the recovery offset: > > > > That would be because the 'Events' number is too old so the 'recovery offset' > > is suspect. > > > > > >> > >> # mdadm -E /dev/sde2 > >> /dev/sde2: > >> Magic : a92b4efc > >> Version : 1.2 > >> Feature Map : 0xb > >> Array UUID : 93e55a39:94681b9d:4c0a1a15:cad51a71 > >> Name : zotac:1 (local to host zotac) > >> Creation Time : Tue Nov 20 08:43:35 2012 > >> Raid Level : raid5 > >> Raid Devices : 4 > >> > >> Avail Dev Size : 3592190757 (1712.89 GiB 1839.20 GB) > >> Array Size : 3918320640 (3736.80 GiB 4012.36 GB) > >> Used Dev Size : 2612213760 (1245.60 GiB 1337.45 GB) > >> Data Offset : 262144 sectors > >> Super Offset : 8 sectors > >> Recovery Offset : 1634547480 sectors > >> State : clean > >> Device UUID : 08019409:984c2217:21706201:039b6bbf > >> > >> Internal Bitmap : 8 sectors from superblock > >> Update Time : Sun Nov 25 10:04:06 2012 > >> Bad Block Log : 512 entries available at offset 262128 sectors > >> Checksum : bd6c4947 - correct > >> Events : 87705 > >> > >> Layout : left-symmetric > >> Chunk Size : 512K > >> > >> Device Role : Active device 3 > >> Array State : AAAA ('A' == active, '.' == missing, 'R' == replacing) > >> > >> > >> > >> 3) What's the correct procedure that allows me recover most of the > >> data, ignoring the readerrors ? > >> > >> 4) I know that sdd2 has read errors at sectors 1634863440 - > >> 1634863800, if i write zero directly to those sectors using 'dd' , how > >> will raid behave during rebuild? After a write it will probably be > >> able to read it again (i dont care about the data on those sectors) . > > > > If it can read, it will, and will write to the new device. > > > >> > >> 5) Right now i am not using replaceable option , but where replaceable > >> option could help me here ? > > > > I don't think 'replaceable' is particularly relevant here. > > > > NeilBrown > > > > > >> > >> > >> Thanks ! > > > > I am still doing some tests and i will report back. > btw, is there anyway to see how many badblock the disk has ? I can see > only the 'bad blocks present.' , but not how many. When the array is active you can look in /sys/block/mdXX/md/dev-YYY/badblocks It will list them all. Currently mdadm doesn't extract this information. I should probably add an --examine-badblocks option to list them. NeilBrown
Attachment:
signature.asc
Description: PGP signature
