Daniel Rock wrote:> Hi,
>
> I just triggered a kernel panic while trying to import a zpool.
>
> The disk in the zpool was residing on a Symmetrix and mirrored with SRDF.
The host sees both devices though (one writeable device "R1" on one
> Symmetrix box and one write protected device "R2" on another
Symmetrix box).
>
> It seems zfs tries to import the write protected device (instead of the
writeable one) and panics.
>
> On VxVM I usually can just import such a disk group with no problems:
> 1. I could mark the write protected disks as "offline":
> vxdisk offline cXtYdZ
> 2. Even if I don''t mark the write protected disks as offline I am
still be able to import the disk group, because VxVM notices the write protected
disks by itself.
>
> Shouldn''t have ZFS a similar mechanism?
>
> The panic only occurs during manual import. On auto-import during bootup
the pool is imported just fine.
>
Yes, ZFS should be able to deal with this... but it obviously doesn''t
yet. This only occurs on an import because we recreate the pool by
probing out the available luns. We apparently have chosen to use some
of the read-only interfaces. This does not happen when we work from the
zpool.cache, since we have all the devices we are going to use for the
pool already listed in the cache, and so don''t attempt to probe.
This will be at least partially addressed when bug 6470435 is fixed.
>
> Some info from the crash dump:
>
>
>>::status
>
> debugging crash dump vmcore.0 (64-bit) from wesson
> operating system: 5.10 Generic_118833-22 (sun4us)
> panic message:
> ZFS: I/O failure (write on <unknown> off 0: zio 300081140c0 [L0
unallocated] 4000L/600P DVA[0]=<0:1400:600> DVA[1]=<0:68001400:600>
fletcher4 lzjb BE contiguous birth=7 fill=0 cksum=6c81986e6d:63188e
> dump content: kernel pages only
>
>>$<msgbuf
>
> [...]
>
> WARNING: /scsi_vhci/ssd at g60060480000287460610424356323645 (ssd38):
> Error for Command: write(10) Error Level: Fatal
> Requested Block: 8236 Error Block: 8236
> Vendor: EMC Serial Number: 1026E000e
> Sense Key: Write Protected
> ASC: 0x27 (write protected), ASCQ: 0x0, FRU: 0x0
> WARNING: /scsi_vhci/ssd at g60060480000xxxxxxxxx424356323645 (ssd38):
> Error for Command: write(10) Error Level: Fatal
> Requested Block: 3416108 Error Block: 3416108
> Vendor: EMC Serial Number: 1026E000e
> Sense Key: Write Protected
> ASC: 0x27 (write protected), ASCQ: 0x0, FRU: 0x0
> [above WARNING pair in total 4 times]
>
> panic[cpu16]/thread=2a101111cc0:
> ZFS: I/O failure (write on <unknown> off 0: zio 300081140c0 [L0
unallocated] 4000L/600P DVA[0]=<0:1400:600> DVA[1]=<0:68001400:600>
fletcher4 lzjb BE contiguous birth=7 fill=0
cksum=6c81986e6d:63188e9e5462:30ab176f69d04a:110eda2de2c1bde2): error 5
>
>
> 000002a101111740 zfs:zio_done+284 (300081140c0, 0, a8, 7019dbf0, 0,
3000b1d4dc0)
> %l0-3: 00000300080c3200 000000007019d800 0000000000000005
0000000000000005
> %l4-7: 000000007b712278 0000000000000002 0000000000000007
0000000000000005
> 000002a101111940 zfs:zio_vdev_io_assess+178 (300081140c0, 8000, 10, 0, 0,
10)
> %l0-3: 0000000000000002 0000000000000002 0000000000000000
0000000000000005
> %l4-7: 0000000000000010 000000000025f1ac 0000000000000000
000000344fdd3258
> 000002a101111a00 genunix:taskq_thread+1a4 (300042dc578, 300042dc520, 50001,
345003f564, 2a101111aca, 2a101111ac8)
> %l0-3: 0000000000010000 00000300042dc548 00000300042dc550
00000300042dc552
> %l4-7: 00000300063ef068 0000000000000002 0000000000000000
00000300042dc540
>
> syncing file systems...
>
>
>>$c
>
> vpanic(7b73b3a0, 7019dbf0, 7019dbc0, 7b73a428, 0, 300081140c0)
> zio_done+0x284(300081140c0, 0, a8, 7019dbf0, 0, 3000b1d4dc0)
> zio_vdev_io_assess+0x178(300081140c0, 8000, 10, 0, 0, 10)
> taskq_thread+0x1a4(300042dc578, 300042dc520, 50001, 345003f564,
2a101111aca, 2a101111ac8)
> thread_start+4(300042dc520, 0, baddcafebaddcafe, baddcafebaddcafe,
baddcafebaddcafe, baddcafebaddcafe)
> --
> This messages posted from opensolaris.org
> _______________________________________________
> zfs-code mailing list
> zfs-code at opensolaris.org
> http://opensolaris.org/mailman/listinfo/zfs-code