parted is worthless!

[saturn-vk]

How come parted can't do anything it's ment to do? They say you can expand and extY partition (if you move the end), yet you really cannot do that. I've tried to grow a ext3 partition(which was on the begining of the drive), and it failed miserably. then i changed it to an ext2 partition. still the same. Mind you, the partition was no mounted during the operation.

Filesystem has incompatiable feature enabled



so is parted only good for vfat fs or am I missing something?

and just to help you visualize:

Name Flags Part Type FS Type [Label] Size (MB)
------------------------------------------------------------------------------
hda1 Boot Primary Linux ext2 24939.05
Pri/Log Free Space 12041.81
hda2 Primary Linux swap 1044.62
hda3 Primary Win95 FAT32 (LBA) 3142.06


even though i have a vfat fs, I dont have windows, so I cannot use partition magic.

[mdwatts]

"parted is worthless"

Must be a rant of some sort, so off it goes to /dev/random.

[bwkaz]

The "incompatible feature" that it's talking about could be the has_journal flag in the ext3 superblock. ext3 is just ext2 with that has_journal flag turned on.

If you turn off the has_journal flag (with a tune2fs -O \^has_journal with the partition unmounted), you could probably run parted without that error.

Alternatively, you could probably run parted without that error if you were using any remotely recent version of it -- I haven't seen that error on an ext3 filesystem since before I moved from Mandrake to LFS (LFS 3.0, a either one year or two ago, I can't remember). I currently have version 1.6.3, which understands ext3 with no problems, but some older versions understood it too.

Or, it might be complaining about some other superblock flag setting. What does dumpe2fs /dev/hdb5 | head -n 40 say on the "filesystem features" line (before you turn off the has_journal flag)?

[saturn-vk]

well unless there's a newer version than 1.6.5 than I guess mine would be the newest. And like I said, i've already turned off the journal with the command you specified, thus the ext2 fs(and not ext3). that's why i can only show you the features after the journal removal:
Filesystem features: dir_index filetype sparse_super
I can just assume that those are standard.
parted just doesn't want to work.

[bwkaz]

Nope. :) dir_index is not standard, at least not yet.

You have an htree-enabled kernel, and an htree-enabled e2fsprogs. (htree speeds up access on large directories by storing the dentries in a hash table, IIRC.) Your e2fsprogs turned on htree support either when the FS was created, or at some point after that.

Turn dir_index off with tune2fs (your version of the manpage should tell you how; it's likely done with tune2fs -O ^dir_index, followed by a fsck -f, but I'm not positive, since my e2fsprogs doesn't have that in the manpage yet, it's old).

Did you manually turn on htree on your FS? Or is this just a default partition creation? If it's default, which distro is dumb enough to put htree code into their default kernel? (last I heard on ext3-users, it wasn't quite bug-free in 2.5 yet, let alone any 2.4 kernel...)

Note also that one of the older versions of e2fsprogs (don't remember which exactly; probably 1.32 or 1.33 or something around there) had htree enabled on all filesystems it created. If the htree code wasn't patched into the kernel, though, it just won't be used, so you won't see bugs in it. Maybe this is what happened.

In any case, turning that feature off should help. You could even convert back to ext3, with that new of a version of parted. :)

[saturn-vk]

unfortunately it doesn't

if i turn off dir_index it says that the ext2 (even though it is ext3) partition has a "rather strange" layout.
and i just updated e2progs to a newer version that doesnt make dir_index automatically. I created a test partition and it still reports that the ext2 (even though its 3) has a strange layout.

[bwkaz]

Then I guess, grab version 1.6.3, compile it, and use it instead. It's worked for me whenever I've done anything with it and ext3. Crappy, but there's a chance...

Well, maybe try recreating the journal now that dir_index is off?

Have you fsck'ed it after disabling htree?