[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: ext3 + raid, is resize2fs neccessary?



On Oct 17, 2003  09:13 -0400, Eric Wood wrote:
> I'm still unclear about these questions:
> 1. Do you have to convert ext3 back to ext2 before resizing, or is resize2fs
> "journal-aware"?
> I assume "yes".

Correct.

> 2. If you shrink the ext3/2 fs, will data occuping those tail-end blocks be
> moved, assuming your fs is not full?
> I assume "yes"

Correct.

> 3. If you shrink your fs by a generous margin, create the raid, and then
> re-grow your fs to fill the device again, will resize2fs step on the raid
> superblock?
> I assume "yes" because if the size parameter is blank, resize2fs "will
> default to the size of the partition", therefore stepping no the raid
> superblock.  So I'll have to manually figure out the size without stepping
> on the last block or two.

Incorrect.  If you do the resizing of e.g. /dev/md0, the RAID code shows
the device size as ending before the MD RAID superblock so resize2fs will
be happy.  Otherwise mke2fs on a new MD RAID device would also overwrite
the superblock and you could never use it safely.

> I was just hoping that resize2fs would somehow auto-detect a special RAID
> blocks and Journal blocks and not step on them.

You can set up a test MD RAID device to check - /dev/hdX will be slightly
larger than /dev/mdY, the difference being the size of the MD superblock.

Cheers, Andreas
--
Andreas Dilger
http://sourceforge.net/projects/ext2resize/
http://www-mddsp.enel.ucalgary.ca/People/adilger/




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]