mounting component device of RAID1 with internal bitmap

mounting component device of RAID1 with internal bitmap

[CoolCold]

Hello!
For raid1 array with metadata 0.90 it is possible to mount component
(let's say /dev/sda1) as usual FS, fix smth, /etc/fstab for example,
unmount and then assamble arary. but is this possible for raid1 array
with internal bitmap ? If my guesses are right this may damage array.

-- 
Best regards,
[COOLCOLD-RIPN]

Re: mounting component device of RAID1 with internal bitmap

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 698 bytes --]

On 04/01/2010 01:47 PM, CoolCold wrote:
> Hello!
> For raid1 array with metadata 0.90 it is possible to mount component
> (let's say /dev/sda1) as usual FS, fix smth, /etc/fstab for example,
> unmount and then assamble arary.

At which point you have just corrupted your array.

> but is this possible for raid1 array
> with internal bitmap ? If my guesses are right this may damage array.

Bitmap doesn't matter, you corrupt your array when you do this either way.



-- 
Doug Ledford <dledford@redhat.com>
              GPG KeyID: CFBFF194
	      http://people.redhat.com/dledford

Infiniband specific RPMs available at
	      http://people.redhat.com/dledford/Infiniband


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: mounting component device of RAID1 with internal bitmap

[CoolCold]

On Thu, Apr 1, 2010 at 10:14 PM, Doug Ledford <dledford@redhat.com> wrote:
> On 04/01/2010 01:47 PM, CoolCold wrote:
>> Hello!
>> For raid1 array with metadata 0.90 it is possible to mount component
>> (let's say /dev/sda1) as usual FS, fix smth, /etc/fstab for example,
>> unmount and then assamble arary.
>
> At which point you have just corrupted your array.
I've did such operation several times and it worked fine..I'll do some
checks and report later.

>
>> but is this possible for raid1 array
>> with internal bitmap ? If my guesses are right this may damage array.
>
> Bitmap doesn't matter, you corrupt your array when you do this either way.
>
>
>
> --
> Doug Ledford <dledford@redhat.com>
>              GPG KeyID: CFBFF194
>              http://people.redhat.com/dledford
>
> Infiniband specific RPMs available at
>              http://people.redhat.com/dledford/Infiniband
>
>



-- 
Best regards,
[COOLCOLD-RIPN]
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: mounting component device of RAID1 with internal bitmap

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 1247 bytes --]

On 04/01/2010 02:26 PM, CoolCold wrote:
> On Thu, Apr 1, 2010 at 10:14 PM, Doug Ledford <dledford@redhat.com> wrote:
>> On 04/01/2010 01:47 PM, CoolCold wrote:
>>> Hello!
>>> For raid1 array with metadata 0.90 it is possible to mount component
>>> (let's say /dev/sda1) as usual FS, fix smth, /etc/fstab for example,
>>> unmount and then assamble arary.
>>
>> At which point you have just corrupted your array.
> I've did such operation several times and it worked fine..I'll do some
> checks and report later.

Because you got lucky.  When you do this, you make the different members
of the raid array disagree on the contents of whatever file you
modified.  Depending on which disk the kernel reads from, you will get
either the new or the old contents of the file.  Any time you mount a
component device and modify it, you *must* force a resync from that
component to all other components or you *will* have a corrupt array.
When that corruption will crop up to bite you in the ass is up to random
chance.


-- 
Doug Ledford <dledford@redhat.com>
              GPG KeyID: CFBFF194
	      http://people.redhat.com/dledford

Infiniband specific RPMs available at
	      http://people.redhat.com/dledford/Infiniband


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Fwd: mounting component device of RAID1 with internal bitmap

[CoolCold]

[-- Attachment #1: Type: text/plain, Size: 2102 bytes --]

Just to share this with list.


---------- Forwarded message ----------
From: Doug Ledford <dledford@redhat.com>
Date: Thu, Apr 1, 2010 at 11:34 PM
Subject: Re: mounting component device of RAID1 with internal bitmap
To: CoolCold <coolthecold@gmail.com>


On 04/01/2010 03:25 PM, CoolCold wrote:
> On Thu, Apr 1, 2010 at 10:39 PM, Doug Ledford <dledford@redhat.com> wrote:
>> On 04/01/2010 02:26 PM, CoolCold wrote:
>>> On Thu, Apr 1, 2010 at 10:14 PM, Doug Ledford <dledford@redhat.com> wrote:
>>>> On 04/01/2010 01:47 PM, CoolCold wrote:
>>>>> Hello!
>>>>> For raid1 array with metadata 0.90 it is possible to mount component
>>>>> (let's say /dev/sda1) as usual FS, fix smth, /etc/fstab for example,
>>>>> unmount and then assamble arary.
>>>>
>>>> At which point you have just corrupted your array.
>>> I've did such operation several times and it worked fine..I'll do some
>>> checks and report later.
>>
>> Because you got lucky.  When you do this, you make the different members
>> of the raid array disagree on the contents of whatever file you
>> modified.  Depending on which disk the kernel reads from, you will get
>> either the new or the old contents of the file.  Any time you mount a
>> component device and modify it, you *must* force a resync from that
>> component to all other components or you *will* have a corrupt array.
>> When that corruption will crop up to bite you in the ass is up to random
>> chance.
> Ah, ofcourse recync is ok , i wanted to know will this break _raid_
> (eg erase some metadata) itself, not actual data on top of it. For 0.9
> it is said in man that it stored at the end of device, but what about
> bitmaps? are they at the end too ?

The bitmaps are outside of the regular filesystem area, so they can't be
harmed by mounting the filesystem.


--
Doug Ledford <dledford@redhat.com>
             GPG KeyID: CFBFF194
             http://people.redhat.com/dledford

Infiniband specific RPMs available at
             http://people.redhat.com/dledford/Infiniband




-- 
Best regards,
[COOLCOLD-RIPN]

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]