Date: Mon, 7 Mar 2011 20:25:31 +0100 From: Fabian Keil <freebsd-listen@fabiankeil.de> To: freebsd-fs@freebsd.org Subject: Re: ZFSv28: log_sysevent: type 19 is not implemented Message-ID: <20110307202531.2c90ff5a@r500.local> In-Reply-To: <20110307200634.3c0f92df@r500.local> References: <20110227202957.GD1992@garage.freebsd.pl> <20110228192129.119cac0c@r500.local> <20110307200634.3c0f92df@r500.local>
next in thread | previous in thread | raw e-mail | index | archive | help
--Sig_/CgFsehXtW9zopU6fiNqtLkR
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
Fabian Keil <freebsd-listen@fabiankeil.de> wrote:
> Fabian Keil <freebsd-listen@fabiankeil.de> wrote:
>=20
> > Pawel Jakub Dawidek <pjd@FreeBSD.org> wrote:
> >=20
> > > I just committed ZFSv28 to HEAD.
>=20
> > Anyway, the things I tested so far (zfs/zpool upgrade,
> > delegation, send, receive, snapshot) worked fine.
>=20
> After unintentionally unplugging an USB disk with this
> zpool on it:
>=20
> | fk@r500 ~ $zpool status toshiba
> | pool: toshiba
> | state: ONLINE
> | scan: none requested
> | config:
> |
> | NAME STATE READ WRITE CKSUM
> | toshiba ONLINE 0 0 0
> | label/toshiba.eli ONLINE 0 0 0
> |
> | errors: No known data errors
>=20
> the system became sluggish and /var/log/messages got spammed
> with error messages:
>=20
> Mar 6 21:33:10 r500 kernel: ugen7.2: <TOSHIBA> at usbus7 (disconnected)
> Mar 6 21:33:10 r500 kernel: umass1: at uhub7, port 1, addr 2 (disconnect=
ed)
> Mar 6 21:33:10 r500 kernel: (pass3:umass-sim1:1:0:0): lost device
> Mar 6 21:33:10 r500 kernel: (pass3:umass-sim1:1:0:0): removing device en=
try
> Mar 6 21:33:10 r500 kernel: (da1:umass-sim1:1:0:0): lost device
> Mar 6 21:33:10 r500 kernel: (da1:umass-sim1:1:0:0): Synchronize cache fa=
iled, status =3D=3D 0xa, scsi status =3D=3D 0x0
> Mar 6 21:33:10 r500 kernel: (da1:umass-sim1:1:0:0): removing device entry
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:33:10 r500 last message repeated 50 times
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implementedlog_=
sysevent: type 19 is not im
> Mar 6 21:33:10 r500 kernel: plemented
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:33:10 r500 last message repeated 238 times
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implementedlog_=
sysevent: type 19 is
> Mar 6 21:33:10 r500 kernel: not implemented
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implementedlog_=
sysevent: type 19 is not impleme
> Mar 6 21:33:10 r500 kernel: nted
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:33:10 r500 last message repeated 87 times
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implementedlog_=
sysevent: type 19 is not implemented
> Mar 6 21:33:10 r500 kernel:=20
> Mar 6 21:33:10 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:33:10 r500 last message repeated 47 times
> Mar 6 21:33:11 r500 kernel: type 19 is not implemented
> Mar 6 21:33:11 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:33:11 r500 last message repeated 1527 times
> Mar 6 21:33:11 r500 kernel: log_sysevent: type 19 is not implementedlog_=
sysevent: type 19 is not implemented
> Mar 6 21:33:11 r500 kernel:=20
> Mar 6 21:33:11 r500 kernel: log_sysevent: type 19 is not implemented
>=20
> fk@r500 ~ $zcat /var/log/messages.0.bz2 | grep -c "type 19 is not impleme=
nted"
> 34101
>=20
> At the time of the unplugging, a video was read from the pool.
>=20
> When trying to export the pool, zpool export hung.
>=20
> After rebooting the system, the message was shown two
> more times between the creation of the provider for the
> main zpool and the swap device:
>=20
> Mar 6 21:43:20 r500 kernel: GEOM_ELI: Device ada0s1d.eli created.
> Mar 6 21:43:20 r500 kernel: GEOM_ELI: Encryption: AES-CBC 128
> Mar 6 21:43:20 r500 kernel: GEOM_ELI: Crypto: software
> Mar 6 21:43:20 r500 kernel: Trying to mount root from ufs:/dev/ada0s1a [=
rw]...
> Mar 6 21:43:20 r500 kernel: WARNING: / was not properly dismounted
> Mar 6 21:43:20 r500 kernel: start_init: trying /sbin/init
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): Requesting SCSI sense d=
ata
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): READ CAPACITY. CDB: 25 =
0 0 0 0 0 0 0 0 0=20
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): CAM status: SCSI Status=
Error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status: Check Cond=
ition
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI sense: NOT READY a=
sc:3a,1 (Medium not present - tray closed)
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): Error 6, Unretryable er=
ror
> Mar 6 21:43:20 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:43:20 r500 kernel: log_sysevent: type 19 is not implemented
> Mar 6 21:43:20 r500 kernel: GEOM_ELI: Device ada0s1b.eli created.
> Mar 6 21:43:20 r500 kernel: GEOM_ELI: Encryption: AES-XTS 256
> Mar 6 21:43:20 r500 kernel: GEOM_ELI: Crypto: software
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): Requesting SCSI sense d=
ata
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): READ CAPACITY. CDB: 25 =
0 0 0 0 0 0 0 0 0=20
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): CAM status: SCSI Status=
Error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status: Check Cond=
ition
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI sense: NOT READY a=
sc:3a,1 (Medium not present - tray closed)
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): Error 6, Unretryable er=
ror
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): Requesting SCSI sense d=
ata
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): READ CAPACITY. CDB: 25 =
0 0 0 0 0 0 0 0 0=20
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): CAM status: SCSI Status=
Error
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI status: Check Cond=
ition
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): SCSI sense: NOT READY a=
sc:3a,1 (Medium not present - tray closed)
> Mar 6 21:43:20 r500 kernel: (cd0:ahcich1:0:0:0): Error 6, Unretryable er=
ror
> Mar 6 21:43:20 r500 kernel: lo1: bpf attached
> Mar 6 21:43:20 r500 kernel: wlan0: bpf attached
> Mar 6 21:43:20 r500 kernel: wlan0: bpf attached
> Mar 6 21:43:20 r500 kernel: wlan0: Ethernet address: 00:[...]
> Mar 6 21:43:20 r500 kernel: firmware: 'iwn5000fw' version 0: 353240 byte=
s loaded at 0xffffffff814120b0
> Mar 6 21:43:20 r500 kernel: firmware: 'iwn5000fw' version 0: 353240 byte=
s loaded at 0xffffffff814120b0
> Mar 6 21:43:20 r500 kernel: bge0: Disabling fastboot
> Mar 6 21:43:20 r500 kernel: bge0: Disabling fastboot
> Mar 6 21:43:20 r500 savecore: /dev/ada0s1b: Operation not permitted
> Mar 6 21:43:21 r500 named[2219]: starting BIND 9.6.3 -t /var/named -u bi=
nd
> Mar 6 21:43:21 r500 named[2219]: built with '--prefix=3D/usr' '--infodir=
=3D/usr/share/info' '--mandir=3D/usr/share/man' '--enable-threads' '--enabl=
e-getifaddrs' '--disable-linux-caps' '--with-openssl=3D/usr' '--with-random=
dev=3D/dev/random' '--without-idn' '--without-libxml2'
> Mar 6 21:43:21 r500 named[2219]: command channel listening on 127.0.0.1#=
953
> Mar 6 21:43:21 r500 named[2219]: command channel listening on ::1#953
> Mar 6 21:43:21 r500 named[2219]: the working directory is not writable
> Mar 6 21:43:21 r500 named[2219]: running
> Mar 6 21:43:53 r500 wpa_supplicant[512]: WPA: Group rekeying completed w=
ith 00:[...] [GTK=3DTKIP]
> Mar 6 21:44:14 r500 syslogd: exiting on signal 15
>=20
> As the boot process got stuck with no additional messages
> printed, I rebooted into single-user mode, exported the
> faulted pool and finished the boot process. The system
> came back normally and the pool could be imported without
> issues
>=20
> fk@r500 ~ $grep zfs /boot/loader.conf | grep -v "^ *#"
> zfs_load=3D"YES"
>=20
> I used the attached patch to stop the log spam, but the
> main issue seems to be reproducible. The top output after
> detaching the pool:
>=20
> last pid: 4985; load averages: 10.47, 3.96, 2.02 up 0+02:01:49 19:=
20:49
> 552 processes: 12 running, 518 sleeping, 22 waiting
> CPU: 1.2% user, 0.0% nice, 98.8% system, 0.0% interrupt, 0.0% idle
> Mem: 267M Active, 95M Inact, 859M Wired, 2032K Cache, 7872K Buf, 692M Free
> Swap: 2048M Total, 2048M Free
>=20
> PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMM=
AND =
=20
> 2 root 1 -8 - 0K 16K CPU1 1 1:26 96.97% g_ev=
ent
> 0 root 387 -8 0 0K 6192K - 0 2:11 80.66% kern=
el
> 1702 root 1 76 0 6276K 796K RUN 0 0:15 11.96% devd
> 11 root 2 155 ki31 0K 32K RUN 0 26:46 0.00% idle
> 3395 fk 1 22 0 465M 329M select 0 7:27 0.00% Xorg
> 3398 fk 1 20 0 96120K 9704K RUN 0 0:49 0.00% e16
> 12 root 22 -84 - 0K 352K WAIT 1 0:37 0.00% intr
> 26 root 1 20 - 0K 16K geli:w 0 0:31 0.00% g_el=
i[0] ada0s1d
> 27 root 1 22 - 0K 16K geli:w 1 0:29 0.00% g_el=
i[1] ada0s1d
>=20
> The stuck zpool's stack:
>=20
> fk@r500 ~ $sudo procstat -kk $(pgrep zpool)
> PID TID COMM TDNAME KSTACK =
=20
> 5087 100490 zpool initial thread mi_switch+0x174 sleepq_wai=
t+0x42 __lockmgr_args+0x7a3 vop_stdlock+0x39 VOP_LOCK1_APV+0x52 _vn_lock+0x=
47 vflush+0x125 zfs_umount+0x9f dounmount+0x31e unmount+0x38b syscallenter+=
0x331 syscall+0x4b Xfast_syscall+0xdd=20
>=20
> When I re-attached the disk, g_event kept eating cpu.
>=20
> I'm using a script to attach and import pools on USB devices and as
> it currently doesn't handle faulted pools, it tried to import the
> already faulted pool, which resulted in a zpool core dump.
>=20
> Mar 7 19:27:25 r500 sudo: fk : TTY=3Dttyv0 ; PWD=3D/home/fk ; USER=
=3Droot ; COMMAND=3D/sbin/geli attach -j - -k /home/fk/geli-keys/toshiba.ke=
y /dev/label/toshiba
> Mar 7 19:27:33 r500 sudo: fk : TTY=3Dttyv0 ; PWD=3D/home/fk ; USER=
=3Droot ; COMMAND=3D/sbin/zpool import toshiba
> Mar 7 19:27:33 r500 kernel: GEOM_ELI: Device label/toshiba.eli created.
> Mar 7 19:27:33 r500 kernel: GEOM_ELI: Encryption: AES-CBC 128
> Mar 7 19:27:33 r500 kernel: GEOM_ELI: Crypto: software
> Mar 7 19:27:33 r500 kernel: pid 5206 (zpool), uid 0: exited on signal 11=
(core dumped)
>=20
> fk@r500 ~ $gdb /sbin/zpool zpool.core=20
> GNU gdb 6.1.1 [FreeBSD]
> Copyright 2004 Free Software Foundation, Inc.
> [...]
> Loaded symbols for /libexec/ld-elf.so.1
> #0 0x000000080085ec7d in avl_insert () from /lib/libavl.so.2
> [New Thread 802807400 (LWP 100524/initial thread)]
> (gdb) where
> #0 0x000000080085ec7d in avl_insert () from /lib/libavl.so.2
> #1 0x000000080085ed5e in avl_add () from /lib/libavl.so.2
> #2 0x0000000801ae9105 in zpool_find_import_cached () from /lib/libzfs.so=
.2
> #3 0x0000000000409deb in zpool_do_import ()
> #4 0x0000000000406fa9 in main ()
>=20
> This time rebooting into single-user mode to export the faulted
> pool wasn't necessary, but I doubt that the patch had anything
> to do with it.
>=20
> The second time I was using today's CURRENT instead of yesterday's,
> but I don't think there were any ZFS-related commits.
I just tried it a third time, this time with the pool imported
but not in active use. Again I couldn't export the pool afterwards:
fk@r500 ~ $sudo zpool list
NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT
tank 228G 166G 62.1G 72% 1.00x ONLINE -
toshiba 1.36T 1.31T 47.9G 96% 1.00x UNAVAIL -
fk@r500 ~ $sudo zpool export toshiba
load: 1.05 cmd: zpool 4117 [tx->tx_sync_done_cv)] 249.93r 0.00u 0.06s 0% 2=
528k
load: 1.05 cmd: zpool 4117 [tx->tx_sync_done_cv)] 250.11r 0.00u 0.06s 0% 2=
528k
load: 1.05 cmd: zpool 4117 [tx->tx_sync_done_cv)] 250.25r 0.00u 0.06s 0% 2=
528k
fk@r500 ~ $sudo procstat -kk $(pgrep zpool)
PID TID COMM TDNAME KSTACK =
=20
4117 102251 zpool initial thread mi_switch+0x174 sleepq_wait+=
0x42 _cv_wait+0x129 txg_wait_synced+0x85 dmu_tx_assign+0x170 spa_history_lo=
g+0x43 zfs_log_history+0x82 zfs_ioc_pool_export+0x2a zfsdev_ioctl+0xe6 devf=
s_ioctl_f+0x7b kern_ioctl+0x102 ioctl+0xfd syscallenter+0x331 syscall+0x4b =
Xfast_syscall+0xdd=20
And importing the pool again wasn't possible either:
fk@r500 ~ $zpool list
NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT
tank 228G 166G 62.1G 72% 1.00x ONLINE -
toshiba 1.36T 1.31T 47.9G 96% 1.00x UNAVAIL -
fk@r500 ~ $sudo zpool import toshiba
cannot import 'toshiba': a pool with that name is already created/imported,
and no additional pools with that name were found
The system stayed responsive, though, and other pools could
still be imported and exported. It also didn't result in a
"log_sysevent: type 19 is not implemented" message.
Fabian
--Sig_/CgFsehXtW9zopU6fiNqtLkR
Content-Type: application/pgp-signature; name=signature.asc
Content-Disposition: attachment; filename=signature.asc
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.17 (FreeBSD)
iEYEARECAAYFAk11MS4ACgkQBYqIVf93VJ2AagCgj3v2JqFg5PX/mIIwqiPwRWbV
LUcAoJExtK7u/bRsdssIvV/F9PzFFhNq
=A/dc
-----END PGP SIGNATURE-----
--Sig_/CgFsehXtW9zopU6fiNqtLkR--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110307202531.2c90ff5a>