|
Hello,
Dne Sunday 12 August 2012 23:09:35, Ben Hutchings napsal(a): > On Thu, 2012-08-09 at 09:40 +0200, root wrote: > > Package: linux-image-3.2.0-3-amd64 > > Version: 3.2.21-3 > > Severity: normal > > > > Hello, > > i'm using this virtual machine to export iscsi targets for windows > > machines to backup. It has 620GB btrfs partition with 550GB iscsi file_io > > LUN exported. I'm using btrfs snapshots to transport (consistent copy of) > > this big file to secondary backup storage (qnap and usb disk) - it takes > > around 12 hours to transport - thats reason to use snapshots. Everything > > was working fine, until i run out of space on btrfs (because of snapshot > > and fullbackup from windows combination).> > > Since than, i have created fresh filesystem and create empty new file to > > export over iscsi.> > > Now, around 250GB of writen data do iscsi file, kern.log starts to fill > > with these messages vfs_writev() returned -28 > > > > It keeps to show on every access to iscsi (even "rescan disk" operation in > > windows triggers it) > Any read access may change the image file's access time (atime) if you > don't use the 'noatime' mount option. On btrfs, metadata changes may > allocate more disk space. > You are right, i forgot noatime option on mount
> > I was suspecting btrfs, but there is no oops in kern.log. > > > > I tried to google this error and found nothing relevant. > > Error 28 is ENOSPC: No space left on device. > > > I tried to reset everything (new filesystem, empty file, remove/add target > > from windows) and this errors comes back after writing approx. 250GB of > > data. (Before problem, file was filled to around 540GB). > > > > Size now: > > # du -cms iscsiSBS > > 244561 iscsiSBS > > [...] > > What does 'btrfs filesystem df /btrfs' say (substitute the actual > mountpoint for the btrfs volume)? It said (i saved this before running balance) Data: total=578.01GB, used=263.03GB System, DUP: total=8.00MB, used=72.00KB System: total=4.00MB, used=0.00 Metadata, DUP: total=1.73GB, used=1.03GB Metadata: total=8.00MB, used=0.00
I tried this - is it ok, it doesn't free "used" after rm?
# dd if=/dev/zero of=/mnt/tmp/1 bs=100M count=10 # rm /mnt/tmp/1 # btrfs filesystem df /mnt/tmp/ Data: total=1.01GB, used=0.00 System, DUP: total=8.00MB, used=4.00KB System: total=4.00MB, used=0.00 Metadata, DUP: total=1.00GB, used=1.27MB Metadata: total=8.00MB, used=0.00
# dd if=/mnt/backupWin/iscsi/iscsiSBS of=/mnt/tmp/1 bs=100M count=10 10+0 records in 10+0 records out 1048576000 bytes (1.0 GB) copied, 22.0851 s, 47.5 MB/s
# btrfs filesystem df /mnt/tmp/ Data: total=1.01GB, used=1000.00MB System, DUP: total=8.00MB, used=4.00KB System: total=4.00MB, used=0.00 Metadata, DUP: total=1.00GB, used=1.41MB Metadata: total=8.00MB, used=0.00
# rm /mnt/tmp/1 # btrfs filesystem df /mnt/tmp/ Data: total=1.01GB, used=1000.00MB System, DUP: total=8.00MB, used=4.00KB System: total=4.00MB, used=0.00 Metadata, DUP: total=1.00GB, used=1.41MB Metadata: total=8.00MB, used=0.00
# dd if=/mnt/backupWin/iscsi/iscsiSBS of=/mnt/tmp/1 bs=100M count=90 dd: writing `/mnt/tmp/1': No space left on device 82+0 records in 81+0 records out 8560574464 bytes (8.6 GB) copied, 280.973 s, 30.5 MB/s
# btrfs filesystem df /mnt/tmp/ Data: total=7.97GB, used=7.97GB System, DUP: total=8.00MB, used=4.00KB System: total=4.00MB, used=0.00 Metadata, DUP: total=1.00GB, used=11.39MB Metadata: total=8.00MB, used=0.00
# rm /mnt/tmp/1 # btrfs filesystem df /mnt/tmp/ Data: total=7.97GB, used=5.65GB System, DUP: total=8.00MB, used=4.00KB System: total=4.00MB, used=0.00 Metadata, DUP: total=1.00GB, used=8.06MB Metadata: total=8.00MB, used=0.00
> > Ben.
Libor |
Attachment:
signature.asc
Description: This is a digitally signed message part.