Drive replace question

Drive replace question

[Wakko Warner]

When a drive is being replaced and has lots of errors, does it get kicked
before the resync is completed?

I'm performing this action now on a system with kernel 3.14.6.  I don't
believe the mdadm version makes a difference but it's debian version 3.3-2

Here's the mdstat for that array:
md1 : active raid1 sde[3](R) sdj[2] sdk[1]
      1465138448 blocks super 1.0 [2/2] [UU]
      [>....................]  recovery =  0.0% (419648/1465138448) finish=174.5min speed=139882K/sec
      bitmap: 3/11 pages [12KB], 65536KB chunk

sdj is being replaced by sde.  Currently, md1 is not in use.

-- 
 Microsoft has beaten Volkswagen's world record.  Volkswagen only created 22
 million bugs.

Re: Drive replace question

[NeilBrown]

[-- Attachment #1: Type: text/plain, Size: 1008 bytes --]

On Fri, 11 Jul 2014 13:13:38 -0400 Wakko Warner <wakko@animx.eu.org> wrote:

> When a drive is being replaced and has lots of errors, does it get kicked
> before the resync is completed?

If it doesn't have a bad block log, then "yes".
If it does, then "no, as long as  it doesn't have too many error".

> 
> I'm performing this action now on a system with kernel 3.14.6.  I don't
> believe the mdadm version makes a difference but it's debian version 3.3-2

The version of mdadm that created the array does make a difference.  Not all
versions create a bad block log.  3.3 and later will create a bad block log.

NeilBrown


> 
> Here's the mdstat for that array:
> md1 : active raid1 sde[3](R) sdj[2] sdk[1]
>       1465138448 blocks super 1.0 [2/2] [UU]
>       [>....................]  recovery =  0.0% (419648/1465138448) finish=174.5min speed=139882K/sec
>       bitmap: 3/11 pages [12KB], 65536KB chunk
> 
> sdj is being replaced by sde.  Currently, md1 is not in use.
> 


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 828 bytes --]