Hi Stefan,
If you run commit e163b66985fa162bd7c1e29a049b336bc2790a2e then you
already have the patch included and ESXi5 should work. But the patch
does not entirely suppress WRITE_SAME errors. Before fallback to
non-accelerated zeroing, ESXi tries WRITE_SAME at least once and
probably re-tries it from time to time. I see these occasional
WRITE_SAME errors too, but definitely not for every zeroing SCSI cmd.
The same holds for "MODE SENSE: unimplemented page/subpage: 0x1c/0x02"
error. If I remember this page has something to do with thin
provisioning, SCSI guys should know better.
It would be fine to get an approval from VMware engineers that this
behavior is really correct ;-)
Anyway, I see no performance or data integrity impact of these dmesg
errors in my vSphere 5 lab setup.
Martin
Dne 3.4.2012 1:29, Nicholas A. Bellinger napsal(a):
On Tue, 2012-03-27 at 14:26 +0200, Stefan Gourguis wrote:
Dear List,
i have connected an new ESXi5 (Update1 Build) to an Debian 6 LIO Target (Git commit e163b66985fa162bd7c1e29a049b336bc2790a2e).
The Esxi Server is single FC connected to Lio (P2P)..
Lio-Target serves n 100 GB LVM Disk via iblock to the ESXi.
I started installing an Windows 7 VM on this Datastore (VMFS 5.54)
"Strong as a LIO-N" gives me a lot of Errors in dmesg:
[ 1146.241497] MODE SENSE: unimplemented page/subpage: 0x1c/0x02
[ 1231.742035] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1231.753493] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1231.765680] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1300.303536] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1300.308241] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1300.312904] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1313.008761] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1313.013449] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1313.018111] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1316.343687] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1316.348386] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1316.353041] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1320.468975] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1320.473646] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
[ 1320.478306] WRITE_SAME w/o UNMAP bit not supported for Block Discard Emulation
...
Hi Stefan,
The bugfix to make this work by default w/o the extra esxcli call to
explicitly disable VAAI on all LUNs was reported by Martin Svec, and
went into v3.2-rc5 here:
target: Fix unsupported WRITE_SAME sense payload
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=67236c44741e250199ccd77f1115568e68cf8848
However it appears I missed the CC to stable on this one..
Greg-KH (CC'ed) would you mind including this into v3.1 and v3.0 stable
releases for distros to pick up..?
Thanks,
--nab
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
