UFS panic on Google Compute Engine

[Julian Stecklina]

On 01/09/2014 03:48 AM, Julian Stecklina wrote:
> I checked the local copy of the image I uploaded to GCE and its
> filesystem is fine. Must be happened during transfer to GCE. Weird.

Okay, I've hit this now the third time. Massive file system corruption.
fsck reports lots of

UNALLOCATED  I=6350  OWNER=root MODE=0
SIZE=0 MTIME=Jan  1 00:00 1970
NAME=/usr/ports/x11-themes/metacity-theme-microgui

UNEXPECTED SOFT UPDATE INCONSISTENCY

and

LINK COUNT DIR I=1175282  OWNER=root MODE=40755
SIZE=512 MTIME=Jan  9 03:28 2014  COUNT 3 SHOULD BE 2
ADJUST? yes

LINK COUNT DIR I=1175287  OWNER=root MODE=40755
SIZE=512 MTIME=Jan  9 03:28 2014  COUNT 3 SHOULD BE 2
ADJUST? yes

It seems rather easy to reproduce. Upload a disk image to GCE and
extract ports. I did it via portsnap fetch extract. I am not sure what
could go so horribly wrong. In two of three cases the filesystem was
beyond repair.

As for problem causes: Has the virtio scsi driver changed significantly
between FreeBSD 9 and 10?

Also weird is that the same virtual disk pops up two times:

da0 at vtscsi0 bus 0 scbus0 target 1 lun 0
da0: <Google PersistentDisk 1> Fixed Direct Access SCSI-6 device
da0: 300.000MB/s transfers
da0: 10240MB (20971520 512 byte sectors: 255H 63S/T 1305C)
da1 at vtscsi0 bus 0 scbus0 target 1 lun 1
da1: <Google PersistentDisk 1> Fixed Direct Access SCSI-6 device
da1: 300.000MB/s transfers
da1: 10240MB (20971520 512 byte sectors: 255H 63S/T 1305C)

Julian