Arch Linux

ghost

Member

Registered: 2011-11-10

Posts: 44

MDADM grow: no progress

Hi,
I'm trying to grow my raid 6.
But the reshape is not working right.

$ mdadm --grow --raid-devices=6 /dev/md0

Returned a message that it needs a Backup file, but the reshape started anyways.

$  cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4] 
md0 : active raid6 sdc1[5] sdc2[6] sdd[4] sde3[7] sdb2[9] sdb1[8]
      2929967616 blocks super 1.2 level 6, 512k chunk, algorithm 2 [6/6] [UUUUUU]
      [>....................]  reshape =  0.0% (0/976655872) finish=459.0min speed=35443K/sec

There is no progress. Any recommendations What I should do?
Is it possible to safely abort this operation?

ghost

Member

Registered: 2011-11-10

Posts: 44

Re: MDADM grow: no progress

Does nobody have any idea?

The reshape is still running…

Personalities : [raid6] [raid5] [raid4] 
md0 : active raid6 sdc1[5] sdc2[6] sdd[4] sde3[7] sdb2[9] sdb1[8]
      2929967616 blocks super 1.2 level 6, 512k chunk, algorithm 2 [6/6] [UUUUUU]
      [>....................]  reshape =  0.0% (0/976655872) finish=46282.2min speed=351K/sec

frostschutz

Member

Registered: 2013-11-15

Posts: 1,647

Re: MDADM grow: no progress

Anything in dmesg?

Did you set strange values for sync min/max speeds?

You have /dev/sdc and /dev/sdb twice in there, which is insane, as it effectively leaves you with no redundancy at all.

Last edited by frostschutz (2015-04-16 10:07:11)

ghost

Member

Registered: 2011-11-10

Posts: 44

Re: MDADM grow: no progress

Anything in dmesg?

Not really, the last entries are:

1445.779880] RAID conf printout:
[ 1445.779887]  --- level:6 rd:5 wd:5
[ 1445.779890]  disk 0, o:1, dev:sdd
[ 1445.779893]  disk 1, o:1, dev:sde3
[ 1445.779895]  disk 2, o:1, dev:sdb1
[ 1445.779898]  disk 3, o:1, dev:sdc2
[ 1445.779900]  disk 4, o:1, dev:sdc1
[ 1445.840462] RAID conf printout:
[ 1445.840466]  --- level:6 rd:5 wd:5
[ 1445.840468]  disk 0, o:1, dev:sdd
[ 1445.840471]  disk 1, o:1, dev:sde3
[ 1445.840473]  disk 2, o:1, dev:sdb1
[ 1445.840476]  disk 3, o:1, dev:sdc2
[ 1445.840478]  disk 4, o:1, dev:sdc1
[ 2752.176412] RAID conf printout:
[ 2752.176418]  --- level:6 rd:6 wd:6
[ 2752.176421]  disk 0, o:1, dev:sdd
[ 2752.176424]  disk 1, o:1, dev:sde3
[ 2752.176426]  disk 2, o:1, dev:sdb1
[ 2752.176429]  disk 3, o:1, dev:sdc2
[ 2752.176431]  disk 4, o:1, dev:sdc1
[ 2752.176433]  disk 5, o:1, dev:sdb2
[ 2752.176483] md: reshape of RAID array md0
[ 2752.176486] md: minimum _guaranteed_  speed: 1000 KB/sec/disk.
[ 2752.176488] md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reshape.
[ 2752.176492] md: using 128k window, over a total of 976655872k.
[57079.092802] md: md0: reshape interrupted.
[57079.573055] md: reshape of RAID array md0
[57079.573061] md: minimum _guaranteed_  speed: 1000 KB/sec/disk.
[57079.573063] md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reshape.
[57079.573069] md: using 128k window, over a total of 976655872k.

The interruption happened, because of some debugging with /sys/block/md0/md/sync_action.

Did you set strange values for sync min/max speeds?

No, dmesg shows normal values (in my oppinion).

You have /dev/sdc and /dev/sdb twice in there, which is insane.

No, I'm using raid6 so I have 2 redundancy disks, this allows me to mix 1TB and 2TB disks.
The performance is in my use case not important.

Last edited by ghost (2015-04-16 13:51:54)

alphaniner

Member

From: Ancapistan

Registered: 2010-07-12

Posts: 2,810

Re: MDADM grow: no progress

You have /dev/sdc and /dev/sdb twice in there, which is insane.

No, I'm using raid6 so I have 2 redundancy disks, this allows me to mix 1TB and 2TB disks.
The performance is in my use case not important.

You understand that losing sdc or sdb would be like losing two disks, and any other drive failure would toast your RAID?

---

But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner

frostschutz

Member

Registered: 2013-11-15

Posts: 1,647

Re: MDADM grow: no progress

No, I'm using raid6 so I have 2 redundancy disks, this allows me to mix 1TB and 2TB disks.

If you want to mix disks of different sizes, you should do separate RAIDs for them. So if you have 1TB, 2TB and 3TB disks, you could create partitions of 1TB size. Then you got a RAID across all disks (using the first 1TB partition). Then another RAID across the 2TB+3TB disks (using the second 1TB partition). Then another RAID across the 3TB disks (using the third 1TB partition).

But you must not use two partitions of the same physical disk in the same RAID. That's just nonsense and will cause you no end of problems. First of all you don't have redundancy that way. If one disk fails, it's gone, since one disk is suddenly representing two members. Also even if you don't particularly care about performance, the speeds will be just horrible since the double member disks always have to do worst case seeks.

I don't know if this is the cause of your growing problem, but it's entirely possible. You're not supposed to use RAID that way.

Backup your data and do it over.

Last edited by frostschutz (2015-04-16 14:18:38)