mount on the same mount point

Em Sexta 09 Outubro 2009 11:49, climb65 escreveu:

> Hello,
>
> I'm trying to mount 2 md devices (/dev/md4 and /dev/md5) on the same
> mount point. I have just read that it is possible.
>
> Note : MD devices are JFS.
>
> But, altough I get no error, I do not get what I expected : the second
> mount command makes the first mount invisible.
>
> So, my 2 questions are :
>
> 1) is it possible to mount 2 MD devices on the same mount point?
> 2) If yes, how?
>
> Thanks

can you read two cds on the same drive?

On Monday 12 October 2009 19:32 in comp.os.linux.misc, somebody
identifying as Joe Beanfish wrote...

> climb65 wrote:
>
>> Hello guys and thanks for your answers.
>>
>> Yes, what I'm looking for is a cumulative mount, as you call it
>> "unionfs".
>>
>> For instance :
>>
>> /dev/md4 and /dev/md5 will be moutned as /mnt/data.
>>
>> In this mount point, I will have a global directory, say GD. On each
>> md device, I have this GD directory.
>>
>> Then, on one md device, I have for example, 3 directories, say DA, DB
>> and DC. On the second md device, I have DD and DE.
>>
>> my resulting mount point would be then as follows :
>>
>> /mnt/data/GD/DA/
>> /mnt/data/GD/DB/
>> /mnt/data/GD/DC/
>> /mnt/data/GD/DD/
>> /mnt/data/GD/DE/
>>
>> But as you said, my potential problem is that in D[A-E] directories,
>> I have files with the same name. There is just the main directory
>> (DA, DB, etc) which differs.
>>
>> Is that a real problem?

As long as you don't end up with identical filenames - case-specific -
in one and the same directory, that should not be a problem.

> Also look into LVM for concatenating volumes into one logical volume,
> RAID0.

Please note that the concatenation of separate volumes into one larger
volume is not the same thing as RAID 0. RAID 0 is a single volume
which is striped across two or more physical disks, so that with two
disks, half of the data - if you have more disks in the stripe, it's
the total amount of data divided by the number of disks - is written to
one disk and the second half to the other disk.

It is called RAID 0 because it doesn't offer redundancy. If one disk in
the array fails, you won't be able to access your data anymore because
a chunk of it will be missing. RAID 0 is typically deployed for
performance reasons, because it can be written to and read from faster
than when using a single disk.

That which you are referring to, i.e. the concatenation of separate
volumes into one larger volume, is not RAID but is typically referred
to as a "linear array". If it is applied to more than one physical
disk, we generally speak of JBOD ("Just a Bunch Of Disks").

--
*Aragorn*
(registered GNU/Linux user #223157)

Aragorn <aragorn.RemoveThis@chatfactory.invalid> writes:

>On Monday 12 October 2009 19:32 in comp.os.linux.misc, somebody
>identifying as Joe Beanfish wrote...

>> climb65 wrote:
>>
>>> Hello guys and thanks for your answers.
>>>
>>> Yes, what I'm looking for is a cumulative mount, as you call it
>>> "unionfs".
>>>
>>> For instance :
>>>
>>> /dev/md4 and /dev/md5 will be moutned as /mnt/data.
>>>
>>> In this mount point, I will have a global directory, say GD. On each
>>> md device, I have this GD directory.
>>>
>>> Then, on one md device, I have for example, 3 directories, say DA, DB
>>> and DC. On the second md device, I have DD and DE.
>>>
>>> my resulting mount point would be then as follows :
>>>
>>> /mnt/data/GD/DA/
>>> /mnt/data/GD/DB/
>>> /mnt/data/GD/DC/
>>> /mnt/data/GD/DD/
>>> /mnt/data/GD/DE/
>>>
>>> But as you said, my potential problem is that in D[A-E] directories,
>>> I have files with the same name. There is just the main directory
>>> (DA, DB, etc) which differs.
>>>
>>> Is that a real problem?

>As long as you don't end up with identical filenames - case-specific -
>in one and the same directory, that should not be a problem.

Identical filenames are fine. Identical filenames and identical paths are the
problem. Ie, if on md4 you have
/DA/carlton.jpg
and on md5 you have /DA/carlton.jpg
then the priority is used to choose one of these to display.
(Not sure what happens if you union mount, and then you erase /DA/carlton.jpg
which erases the priority one, does the other one pop up to take its place?)

On Oct 12, 8:08 pm, Unruh <unruh-s... DeleteThis @physics.ubc.ca> wrote:

>
> Identical filenames are fine. Identical filenames and identical paths are the
> problem. Ie, if on md4 you have
> /DA/carlton.jpg
> and on md5 you have /DA/carlton.jpg
> then the priority is used to choose one of these to display.
> (Not sure what happens if you union mount, and then you erase /DA/carlton..jpg
> which erases the priority one, does the other one pop up to take its place?)

What happens depends on if "whiteouts" are implemented.

If I remember correctly, the Mac OS X implementation shows the
"under" file, but you can't delete it after deleting the "upper",
while Linux won't show the "under" after that upper is deleted - Linux
unionfs implements "whiteouts", the OS X didn't (though it may now - I
haven't tested it recently).

On Tuesday 13 October 2009 02:08 in comp.os.linux.misc, somebody
identifying as Unruh wrote...

> Aragorn <aragorn RemoveThis @chatfactory.invalid> writes:
>
>> On Monday 12 October 2009 19:32 in comp.os.linux.misc, somebody
>> identifying as Joe Beanfish wrote...
>
>>> climb65 wrote:
>>>
>>>> Yes, what I'm looking for is a cumulative mount, as you call it
>>>> "unionfs".
>>>>
>>>> For instance :
>>>>
>>>> /dev/md4 and /dev/md5 will be moutned as /mnt/data.
>>>>
>>>> In this mount point, I will have a global directory, say GD. On
>>>> each md device, I have this GD directory. Then, on one md device, I
>>>> have for example, 3 directories, say DA, DB and DC. On the second
>>>> md device, I have DD and DE.
>>>>
>>>> my resulting mount point would be then as follows :
>>>>
>>>> /mnt/data/GD/DA/
>>>> /mnt/data/GD/DB/
>>>> /mnt/data/GD/DC/
>>>> /mnt/data/GD/DD/
>>>> /mnt/data/GD/DE/
>>>>
>>>> But as you said, my potential problem is that in D[A-E]
>>>> directories, I have files with the same name. There is just the
>>>> main directory (DA, DB, etc) which differs.
>>>>
>>>> Is that a real problem?
>
>> As long as you don't end up with identical filenames -
>> case-specific - in one and the same directory, that should not be a
>> problem.
>
> Identical filenames are fine. Identical filenames and identical paths
> are the problem. Ie, if on md4 you have
> /DA/carlton.jpg
> and on md5 you have /DA/carlton.jpg
> then the priority is used to choose one of these to display.

As I understand it, the one displayed will be the one of the filesystem
listed first. It will cover/hide the one(s) on the filesystem(s)
listed behind that in the union.

> (Not sure what happens if you union mount, and then you erase
> /DA/carlton.jpg which erases the priority one, does the other one pop
> up to take its place?)

Possibly, yes, but - again, as I understand it, from only briefly
looking into this thing myself - this is something that can be
specified at mount time. There should be several mount options there
to the /unionfs/ itself.

A quick Google search leads me to this:

http://myhowtosandprojects.blogspot.com/2008/07/install-and-use-u...nfs-mer

<blockquote>

Writing to union

Merging read-only directories is useful in many cases, but the
union itself remains read-only too, until a read-write branch
is added to it. In that case, all changes are stored in leftmost
branch (using copy-up method, see below) and file deletions are
done by using one of the two methods available:

* WHITEOUT mode, inserts a .wh (whiteout) file to mask out a real
file

* DELETE_ALL mode, tries to delete all instances of a file from
all branches

WHITEOUT mode is used as default. Copy-up is a special method used
to handle file modifications in union. A file from ro branch can't
be modified, so it is copied to upper (left) read-write branch at
the time when the modification should begin. Then the modification
is possible and modified file remains in rw branch.

</blockquote>

--
*Aragorn*
(registered GNU/Linux user #223157)

On Mon, 12 Oct 2009 18:56:58 +0000, Unruh wrote :

> Joe Beanfish <joe.DeleteThis@nospam.duh> writes:
>
>>climb65 wrote:
>>> On Fri, 09 Oct 2009 11:10:13 -0700, Tony Lawrence (pcunix) wrote :
>>> (...)
>>>
>>>
>>>
>>> Hello guys and thanks for your answers.
>>>
>>> Yes, what I'm looking for is a cumulative mount, as you call it
>>> "unionfs".
>>>
>>> For instance :
>>>
>>> /dev/md4 and /dev/md5 will be moutned as /mnt/data.
>>>
>>> In this mount point, I will have a global directory, say GD. On each
>>> md device, I have this GD directory.
>>>
>>> Then, on one md device, I have for example, 3 directories, say DA, DB
>>> and DC. On the second md device, I have DD and DE.
>>>
>>> my resulting mount point would be then as follows :
>>>
>>> /mnt/data/GD/DA/
>>> /mnt/data/GD/DB/
>>> /mnt/data/GD/DC/
>>> /mnt/data/GD/DD/
>>> /mnt/data/GD/DE/
>>>
>>> But as you said, my potential problem is that in D[A-E] directories, I
>>> have files with the same name. There is just the main directory (DA,
>>> DB, etc) which differs.
>
> You give unionfs a list of priorities. Thus if you make /dev/md5 have
> priority, then if there are files DA/file1 on both md4 and md5 it is the
> one on md5 that you will see in the unionfs.
>
> Note that having DA/file1 on md4 and DE/file1 on md5 is not a problem.
> They are different -- are in diffeent directories. Just as you can have
> the same filename in different directories on one filesystem. I do not
> know what you call D[A-E] the "main directory" I assume you want to
> mount both md4 and md5 onto /mnt/data/GD. That is the "main directory"
>
>

That sounds good. My directory D(n) will always be on a unique device.
So, a file DA/file1 will never be both on md4 and md5.

There is just the global directory GD that will be on both md4 and
md5. But inside GD, I will have the D(n) directories, each of them in a
unique device (either md4 or md5)

So, here is a schema of what I could get :
_
/mnt/data/GD/DA/file1 )
/mnt/data/GD/DB/file1 )-- on md4 (RAID 10)
/mnt/data/GD/DC/file1 _)
_
/mnt/data/GD/DD/file1 )__ on md5 (RAID 10)
/mnt/data/GD/DE/file1 _)


Is there any problem with the parent directory GD? I think no but advice
is welcome.

Thx

On Tuesday 13 October 2009 16:09 in comp.os.linux.misc, somebody
identifying as climb65 wrote...

> [...] My directory D(n) will always be on a unique device.
> So, a file DA/file1 will never be both on md4 and md5.
>
> There is just the global directory GD that will be on both md4 and
> md5. But inside GD, I will have the D(n) directories, each of them in
> a unique device (either md4 or md5)
>
> So, here is a schema of what I could get :
> _
> /mnt/data/GD/DA/file1 )
> /mnt/data/GD/DB/file1 )-- on md4 (RAID 10)
> /mnt/data/GD/DC/file1 _)
> _
> /mnt/data/GD/DD/file1 )__ on md5 (RAID 10)
> /mnt/data/GD/DE/file1 _)
>
> Is there any problem with the parent directory GD? I think no but
> advice is welcome.

With the intended layout as you have depicted it here, there should be
no problem whatsoever. You should be able to use /unionfs/ to mount
both devices to */mnt/data/GD.*

--
*Aragorn*
(registered GNU/Linux user #223157)