Journalled quotas?

[capscrew]

@ palo_m,

Yes, I subsequently saw the usrjquota reference. I did a my own Google search. One references was here.

As pcmasher did not reply, I dropped the issue. It seems that there is not much interest in this at this time.

A howto to would be great if and/or when you have the time.

[palo_m]

OK, I couldn't wait. As soon as I came home, I had to try it.
Short summary: it's as I already described.

Long story follows.
Warning: For simplicity I ran those commands under root account, so if you try to repeat the same steps, be careful during typing and think about consequences of your actions. Or use normal user for most of actions and sudo for, only mount/umount and quota-related commands need root access. It's your machine and your responsibility...

1. Create image file for testing
   

   linux:~#
   linux:~# cd /tmp
   linux:/tmp# dd if=/dev/zero of=/tmp/disk.img bs=1024 count=20000

2. Make ext3 filesystem in image file
   

   linux:/tmp# mkfs.ext3 -m 0 -L test /tmp/disk.img
   mke2fs 1.41.3 (12-Oct-2008)
   /tmp/disk.img is not a block special device.
   Proceed anyway? (y,n) y
   ...blah-blah-blah... (usual stuff displayed)

3. Mount disk with non-journaled quota option and create some file there... just to have situation like normally used filesystem
   

   linux:/tmp# mount -o loop,rw,usrquota /tmp/disk.img /mnt
   linux:/tmp# mount | grep mnt
   /tmp/disk.img on /mnt type ext3 (rw,loop=/dev/loop0,usrquota)
   linux:/tmp# mkdir /mnt/test1
   linux:/tmp# dd if=/dev/zero of=/mnt/test1/test2 bs=1024 count=100
   100+0 records in
   100+0 records out
   102400 bytes (102 kB) copied, 0.000684922 s, 150 MB/s
   linux:/tmp#
   linux:/tmp#
   linux:/tmp# ls -al /mnt/test1/
   total 103
   drwxr-xr-x 2 root root   1024 2009-04-15 21:30 .
   drwxr-xr-x 4 root root   1024 2009-04-15 21:29 ..
   -rw-r--r-- 1 root root 102400 2009-04-15 21:30 test2
   linux:/tmp# df /mnt
   Filesystem           1K-blocks      Used Available Use% Mounted on
   /tmp/disk.img            19362      1301     18061   7% /mnt

4. Initiate non-journaled quota. It's in vfsv0 format by default, that's what we like:
   

   linux:/tmp# quotacheck -c -v /mnt
   quotacheck: Your kernel probably supports journaled quota but you are not using it. Consider switching to journaled quota to avoid running quotacheck after an unclean shutdown.
   quotacheck: Scanning /dev/loop0 [/mnt] done
   quotacheck: Cannot stat old user quota file: No such file or directory
   quotacheck: Old group file not found. Usage will not be substracted.
   quotacheck: Checked 4 directories and 3 files
   quotacheck: Old file not found.
   linux:/tmp# setquota -u root 10000 15000 1000 1500 /mnt
   linux:/tmp# repquota /mnt
   *** Report for user quotas on device /dev/loop0
   Block grace time: 7days; Inode grace time: 7days
                           Block limits                File limits
   User            used    soft    hard  grace    used  soft  hard  grace
   ----------------------------------------------------------------------
   root      --    1301   10000   15000              6  1000  1500
   
   
   linux:/tmp# ls -al /mnt
   total 24
   drwxr-xr-x  4 root root  1024 2009-04-15 21:57 .
   drwxr-xr-x 23 root root  4096 2009-04-11 18:02 ..
   -rw-------  1 root root  6144 2009-04-15 21:57 aquota.user
   drwx------  2 root root 12288 2009-04-15 21:28 lost+found
   drwxr-xr-x  2 root root  1024 2009-04-15 21:30 test1
   linux:/tmp# quotacheck -v /mnt
   quotacheck: Your kernel probably supports journaled quota but you are not using it. Consider switching to journaled quota to avoid running quotacheck after an unclean shutdown.
   quotacheck: Scanning /dev/loop0 [/mnt] done
   quotacheck: Old group file not found. Usage will not be substracted.
   quotacheck: Checked 4 directories and 4 files

   Now we are in pretty much same situation on /mnt, as our real FS is - we have some files, we have quota set, all seems fine, but quotacheck -v annoys us, that we should consider journaled quota. So, we considered it already and we want it, so let's go for it.
   -
5. Remount filesystem with journaled quota options. (See above, that file aquota.user already exists. The old quota format - vfsold - uses file quota.user instead. I guess we have no reason to use the old format anymore - and there is convertquota command for case of "historical" disks anyway.)
   

   linux:/tmp# mount -o remount,loop,rw,usrjquota=aquota.user,jqfmt=vfsv0 /mnt
   linux:/tmp# mount | grep mnt
   /tmp/disk.img on /mnt type ext3 (rw,loop=/dev/loop0,usrquota,usrjquota=aquota.user,jqfmt=vfsv0)

   We can see, that previous usrquota option stayed there... Alternatively to remount, we could use umount and then mount - but the filesystem cannot be used to be able to do that, which is not always the case. So I prefer remount.
   -
6. Let's see where we are now
   

   linux:/tmp# quotacheck -v /mnt
   quotacheck: Scanning /dev/loop0 [/mnt] done
   quotacheck: Old group file not found. Usage will not be substracted.
   quotacheck: Checked 4 directories and 4 files
   linux:/tmp# repquota /mnt
   *** Report for user quotas on device /dev/loop0
   Block grace time: 7days; Inode grace time: 7days
                           Block limits                File limits
   User            used    soft    hard  grace    used  soft  hard  grace
   ----------------------------------------------------------------------
   root      --    1301   10000   15000              6  1000  1500

   Yeah, the message has gone. We are using journaled quota now!
   
   
7. So as we tested it works with image file, we can try with real partitions. Turn quota off first, then remount it with additional "usrjquota=aquota.user,jqfmt=vfsv0" options, verify with quotacheck that it works and then turn quota back on. If all seems to work fine, one last step is to update /etc/fstab to have the mount options permanent (and if we want to be really-really sure, we can also reboot afterwards... although another remount using /etc/fstab is sufficient).

I did not try it with grpquota (I don't use group quotas anyway), but I believe it works the same, just needs to use "grpjquota=aquota.group"... Here we go, using journaled quota now

[capscrew]

@ palo_m,

That's great work! I believe you are the only one who has documented this feature.

[cupinzim]

Brilliant! Thanx a lot!

[bfdhud]

Tried this on 9.10 / ext4 both user and group quota's work with his tutorial.


hi5

[PieterN]

Thanks a lot for your howto!

For me to get it to work I needed to add "grpjquota=aquota.group" to the mount options, otherwise I got the message:

mount: / not mounted already, or bad option

I got this error when "grpjquota" was left out on both 9.04 and 9.10.

[Alan-A]

There is a further example of a way to start journaled quotas here: http://www.howtoforge.com/perfect-se...ispconfig-2-p4
Look for point 12 "Journaled Quota".
Hope this may help.
A.