When a user exceeds the disk quota configured for a dataset, the dataset becomes unusable: the user cannot even delete her own files from it to free some disk space. Fix: A workaround: echo . > bigfile This workaround is not helpful unless the user has shell access to the system. How-To-Repeat: $ zfs list -o quota,mountpoint d01/home QUOTA MOUNTPOINT 6G /home $ pwd /home/sudakov $ dd if=/dev/zero of=bigfile bs=1m dd: bigfile: Disc quota exceeded 2869+0 records in 2868+1 records out 3007447040 bytes transferred in 88.749210 secs (33887029 bytes/sec) $ rm bigfile rm: bigfile: Disc quota exceeded $
Responsible Changed From-To: freebsd-bugs->freebsd-fs Over to maintainer(s).
State Changed From-To: open->feedback 1) do you have snapshots? 2) could you try: http://svn.freebsd.org/base/user/kmacy/ZFS_MFC/
Responsible Changed From-To: freebsd-fs->kmacy 1) do you have snapshots? 2) could you try: http://svn.freebsd.org/base/user/kmacy/ZFS_MFC/
kmacy@FreeBSD.org wrote: > Synopsis: [zfs] User cannot delete a file when a ZFS dataset is full. > > State-Changed-From-To: open->feedback > State-Changed-By: kmacy > State-Changed-When: Sun May 17 05:49:18 UTC 2009 > State-Changed-Why: > > > 1) do you have snapshots? No, I don't. [sudakov@vas ~] zfs list NAME USED AVAIL REFER MOUNTPOINT d01 96,7G 9,57G 795K /d01 d01/home 4,71G 3,29G 4,71G /home d01/media 90,4G 9,57G 90,4G /msdos d01/ports 895M 9,57G 895M /usr/ports d01/soft 764M 9,57G 764M /usr/local d01/swap 90K 9,57G 90K - [sudakov@vas ~] dd if=/dev/zero of=bigfile bs=1m dd: bigfile: Disc quota exceeded 3369+0 records in 3368+1 records out 3532128256 bytes transferred in 115.604244 secs (30553621 bytes/sec) [sudakov@vas ~] rm bigfile rm: bigfile: Disc quota exceeded [sudakov@vas ~] > > 2) could you try: http://svn.freebsd.org/base/user/kmacy/ZFS_MFC/ Excuse me, what should I do with this? I could try a patch against 7.1-RELEASE if you give me one. -- Victor Sudakov, VAS4-RIPE, VAS47-RIPN sip:sudakov@sibptus.tomsk.ru
On Sun, May 17, 2009 at 4:50 AM, Victor Sudakov <sudakov@sibptus.tomsk.ru> wrote: > kmacy@FreeBSD.org wrote: >> Synopsis: [zfs] User cannot delete a file when a ZFS dataset is full. >> >> State-Changed-From-To: open->feedback >> State-Changed-By: kmacy >> State-Changed-When: Sun May 17 05:49:18 UTC 2009 >> State-Changed-Why: >> >> >> 1) do you have snapshots? > > No, I don't. > > [sudakov@vas ~] zfs list > NAME =A0 =A0 =A0 =A0USED =A0AVAIL =A0REFER =A0MOUNTPOINT > d01 =A0 =A0 =A0 =A096,7G =A09,57G =A0 795K =A0/d01 > d01/home =A0 4,71G =A03,29G =A04,71G =A0/home > d01/media =A090,4G =A09,57G =A090,4G =A0/msdos > d01/ports =A0 895M =A09,57G =A0 895M =A0/usr/ports > d01/soft =A0 =A0764M =A09,57G =A0 764M =A0/usr/local > d01/swap =A0 =A0 90K =A09,57G =A0 =A090K =A0- > [sudakov@vas ~] dd if=3D/dev/zero of=3Dbigfile bs=3D1m > dd: bigfile: Disc quota exceeded > 3369+0 records in > 3368+1 records out > 3532128256 bytes transferred in 115.604244 secs (30553621 bytes/sec) > [sudakov@vas ~] rm bigfile > rm: bigfile: Disc quota exceeded > [sudakov@vas ~] > >> >> 2) could you try: http://svn.freebsd.org/base/user/kmacy/ZFS_MFC/ > > Excuse me, what should I do with this? > > I could try a patch against 7.1-RELEASE if you give me one. I don't envision a patch for 7.1-RELEASE. This is a MFC to 7-STABLE. FWIW, I am not able to reproduce this in a VM. -Kip
State Changed From-To: feedback->closed ZFS v13 reserves enough space to delete files.
Kip Macy wrote: > FWIW, I am not able to reproduce this in a VM. I hope you were not working as root. I also cannot reproduce the problem when trying as root. The problem can be reproduced only under an ordinary user. -- Victor Sudakov, VAS4-RIPE, VAS47-RIPN sip:sudakov@sibptus.tomsk.ru
No. What happens now is that the dd just becomes infinitely slow and never quite fills up the pool. -Kip On Sun, May 17, 2009 at 9:09 PM, Victor Sudakov <sudakov@sibptus.tomsk.ru> wrote: > Kip Macy wrote: >> FWIW, I am not able to reproduce this in a VM. > > I hope you were not working as root. > I also cannot reproduce the problem when trying as root. The problem > can be reproduced only under an ordinary user. > > -- > Victor Sudakov, =A0VAS4-RIPE, VAS47-RIPN > sip:sudakov@sibptus.tomsk.ru > --=20 When bad men combine, the good must associate; else they will fall one by one, an unpitied sacrifice in a contemptible struggle. Edmund Burke