xfs
[Top] [All Lists]

Re: SAN resizing disk, no LVM

To: daves@xxxxxxxxxxxxxxx
Subject: Re: SAN resizing disk, no LVM
From: Keith Owens <kaos@xxxxxxx>
Date: Thu, 28 Apr 2005 09:47:19 +1000
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: Your message of "Wed, 27 Apr 2005 16:34:38 MST." <4270218E.80708@activestate.com>
Sender: linux-xfs-bounce@xxxxxxxxxxx
On Wed, 27 Apr 2005 16:34:38 -0700, 
David Sparks <daves@xxxxxxxxxxxxxxx> wrote:
>>>Is using /dev/sdb instead of /dev/sdb1 for a fs safe to do?  (I've never
>>>read anything authoritative on this subject)
>> 
>> 
>> A block device is a block device is a ...  It is safe, as long as no
>> other tool tries to access /dev/sdb.  In particular, if anything tries
>> to partition /dev/sdb or to install a boot loader on /dev/sdb then it
>> will overwrite the first superblock of the XFS partition.  You are
>> gaining a little extra space, at the risk that some other disk tool
>> will accidentally destroy your filesystem.
>
>The main idea is that LVM isn't required to resize the disk.  A
>`xfs_growfs /mnt/sdb` is sufficient
>
>Is there a simple way to resize a partition with a fs inside it?

Resizing any partitition by increasing the end address is easy, as long
as there is free space after the parttion being grown.  Both parted and
fdisk can change the end address, then you run xfs_growfs.  Resizing by
changing the start address is hard, all the data has to be moved.

>I do appreciate the warning, I see this every time I load fdisk:
>
> Warning: invalid flag 0x0000 of partition table 4 will be corrected by
>w(rite)
>
>
>I tried to "correct" the invalid flag once and the fs was unmountable.
>It did get fixed from a xfs_repair but I don't know if there was any
>actual damage to the fs?  It is only a test fs at the moment and mostly
>empty so there was no indication of data loss.

Your partition "correction" overwrote the first 512 bytes of the first
XFS superblock.  Fortunately xfs_repair can recover from that condition
by using a backup superblock.


<Prev in Thread] Current Thread [Next in Thread>