[REPORT] ZFS panic on recent 7-current

Ståle Kristoffersen staalebk at ifi.uio.no
Mon Mar 19 03:16:24 UTC 2007


On 2007-03-18 at 14:42, Pawel Jakub Dawidek wrote:
> You can't use recent HEAD, because of tegge's recent changes that break
> API/ABI. You need to have HEAD from before tegge's changes.

Ah, that sorted things out for me as well. I did get a 
"kmem_malloc(X): kmem_map too small: Y total allocated"-panic, is that
still something that should be ignored?

It also looks like 'zpool list' shows the wrong(?) size and available info if
showing a raidz-pool:
(ad1-3 512MB disks)

# zpool create test raidz ad1 ad2 ad3
# zpool list
NAME                    SIZE    USED   AVAIL    CAP  HEALTH     ALTROOT
test                   1.48G    157K   1.48G     0%  ONLINE     -
# df -ht zfs
Filesystem     Size    Used   Avail Capacity  Mounted on
test           980M     32K    980M     0%    /test
# dd if=/dev/random of=/test/temp bs=1M count=512
512+0 records in
512+0 records out
536870912 bytes transferred in 40.188660 secs (13358766 bytes/sec)
# zpool list
NAME                    SIZE    USED   AVAIL    CAP  HEALTH     ALTROOT
test                   1.48G    773M    747M    50%  ONLINE     -
# df -ht zfs
Filesystem     Size    Used   Avail Capacity  Mounted on
test           980M    515M    466M    53%    /test

It could be argued that it shows the correct amount that is consumed with
redundant information, but when using a mirror instead, output is more as
expected:

# zpool destroy test
# zpool create test mirror ad1 ad2
# zpool list
NAME                    SIZE    USED   AVAIL    CAP  HEALTH     ALTROOT
test                    504M     88K    504M     0%  ONLINE     -
# df -ht zfs
Filesystem     Size    Used   Avail Capacity  Mounted on
test           472M     32K    472M     0%    /test

It could be that this is how it should be, but I think it is a bit
confusing, it could also be that it is fixed in perforce, but I couldn't
find anything in the log or found out how to check out a recent copy of the
code.

Thanks for the brilliant work on porting zfs, keep it up!
-- 
Ståle Kristoffersen
staalebk at ifi.uio.no


More information about the freebsd-fs mailing list