[Bug 219972] Unable to zpool export following some zfs recv
bugzilla-noreply at freebsd.org
bugzilla-noreply at freebsd.org
Wed Jun 14 14:23:52 UTC 2017
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219972
--- Comment #1 from pfribeiro at gmail.com ---
I've done some further testing on this by isolating the issue to a 'zfs send'
and 'zfs recv' on the same machine between the external USB drive (on /dev/da0)
and a USB stick (on /dev/da1).
1. I made sure that there were no existing pools on both devices, by destroying
the pools and doing 'zpool labelclear -f /dev/da0' and 'zpool labelclear -f
/dev/da1'.
root at sunflower:~ # zpool labelclear -f /dev/da0
root at sunflower:~ # zpool labelclear -f /dev/da1
2. Created two pools:
root at sunflower:~ # zpool create -O atime=off -O compression=lz4 master /dev/da0
root at sunflower:~ # zpool create -O atime=off -O compression=lz4 slave /dev/da1
root at sunflower:~ # zpool list
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
master 7.25T 360K 7.25T - 0% 0% 1.00x ONLINE -
slave 7.25G 78.5K 7.25G - 0% 0% 1.00x ONLINE -
zroot 51.5G 1.11G 50.4G - 0% 2% 1.00x ONLINE -
3. Created some test file on the master pool:
root at sunflower:~ # touch /master/test
root at sunflower:~ # echo "Hello" > /master/test
4. Created a snapshot on the master pool:
root at sunflower:~ # zfs snapshot -r "master at 1"
5. Exported the 'slave' pool and imported unmounted (not sure this matters
really):
root at sunflower:~ # zpool export slave
root at sunflower:~ # zpool import -N slave
6. Sent the snapshot across to the slave pool:
root at sunflower:~ # zfs send -Rev "master at 1" | zfs recv -Fu slave
full send of master at 1 estimated size is 10K
total estimated size is 10K
TIME SENT SNAPSHOT
root at sunflower:~ # zpool list
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
master 7.25T 588K 7.25T - 0% 0% 1.00x ONLINE -
slave 7.25G 376K 7.25G - 0% 0% 1.00x ONLINE -
zroot 51.5G 1.11G 50.4G - 0% 2% 1.00x ONLINE -
7. Tried to export and import the slave a few times:
root at sunflower:~ # zpool export slave
root at sunflower:~ # zpool import slave
root at sunflower:~ # zpool export slave
cannot export 'slave': pool is busy
...and bang the problem is back. This seems a little crazy.
I would appreciate if someone could try and help me reproduce this problem on
their system, and perhaps point me to some further debugging steps. I'm out of
ideas as to what else I could/should test at this point.
Thanks
--
You are receiving this mail because:
You are the assignee for the bug.
More information about the freebsd-fs
mailing list