LSI SAS 2008 (mfi) on SuperMicro X8SI6-F

[Dmitry Morozovsky]

Dear colleagues,

are there any success stories with using SuperMicro LSI SAS with stable/8 ?

I tried mfi drivers sources from LSI site (had to add one #include to make
kdump compilation happy) with no success

pciconf info is

none@pci0:0:2:0:    class=0x010700 card=0x00721000 chip=0x00721000
rev=0x02 hdr=0x00
    vendor     = 'LSI Logic (Was: Symbios Login, NCR)'
    class      = mass storage
    subclass   = SAS

Thanks in advance!

--
Sincerely,
D.Marck                                     [DM5020, MCK-RIPE, DM3-RIPN]
[ FreeBSD committer:                                 [hidden email] ]
------------------------------------------------------------------------
*** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [hidden email] ***
------------------------------------------------------------------------
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Damien Fleuriot-2]

Hi,


mps0@pci0:1:0:0:        class=0x010700 card=0x1f1d1028 chip=0x00721000
rev=0x02 hdr=0x00
    vendor     = 'LSI Logic (Was: Symbios Logic, NCR)'
    class      = mass storage
    subclass   = SAS
    bar   [10] = type I/O Port, range 32, base 0xfc00, size 256, enabled
    bar   [14] = type Memory, range 64, base 0xdf2b0000, size 65536, enabled
    bar   [1c] = type Memory, range 64, base 0xdf2c0000, size 262144,
enabled
    cap 01[50] = powerspec 3  supports D0 D1 D2 D3  current D0
    cap 10[68] = PCI-Express 2 endpoint max data 256(4096) link x8(x8)
    cap 03[d0] = VPD
    cap 05[a8] = MSI supports 1 message, 64 bit
    cap 11[c0] = MSI-X supports 15 messages in map 0x14 enabled
ecap 0001[100] = AER 1 0 fatal 0 non-fatal 0 corrected
ecap 0004[138] = unknown 1


This is on a h200 card from dell using LSI 2008 on FreeBSD 8.2-RC3 amd64.

Now, the documentation is FAR from finished but you may see a draft here:
http://my.gd/bsd.htm


What you want to do is :
- delete your logical volume on the RAID controller
- boot an MFSBSD image preloaded with a kernel containing the mps(4)
module from 9.0.
- this will present you the disks as /dev/da0 and /dev/da1
- setup a software mirror from there

If you like, you may mail me the network configuration you'd like to get
to be able to reach the host (MAC address of the NIC, IP address, mask,
gateway, nameserver) and your public ssh key, and I'll generate you a
MFSBSD 8.2-RC3 image preloaded with mps(4).

You then boot it via PXE or inject it (install linux, download image,
inject using dd).


On 2/16/11 4:48 PM, Dmitry Morozovsky wrote:

> Dear colleagues,
>
> are there any success stories with using SuperMicro LSI SAS with stable/8 ?
>
> I tried mfi drivers sources from LSI site (had to add one #include to make
> kdump compilation happy) with no success
>
> pciconf info is
>
> none@pci0:0:2:0:    class=0x010700 card=0x00721000 chip=0x00721000
> rev=0x02 hdr=0x00
>     vendor     = 'LSI Logic (Was: Symbios Login, NCR)'
>     class      = mass storage
>     subclass   = SAS
>
> Thanks in advance!
>

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Daniel Kalchev]

I have sucessfully used that motherboard with FreeBSD 9 and the mps
driver. The mfi driver found on the LSI site does not support this
controller.

Daniel

PS: My experiments were with the X8DTL-6F motherboard and Supermicro
chassis with E16 expander. There is no reason the HBA chip in the single
processor motherboard to be different.

On 16.02.11 17:48, Dmitry Morozovsky wrote:

> Dear colleagues,
>
> are there any success stories with using SuperMicro LSI SAS with stable/8 ?
>
> I tried mfi drivers sources from LSI site (had to add one #include to make
> kdump compilation happy) with no success
>
> pciconf info is
>
> none@pci0:0:2:0:    class=0x010700 card=0x00721000 chip=0x00721000
> rev=0x02 hdr=0x00
>      vendor     = 'LSI Logic (Was: Symbios Login, NCR)'
>      class      = mass storage
>      subclass   = SAS
>
> Thanks in advance!
>

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Dmitry Morozovsky]

On Wed, 16 Feb 2011, Daniel Kalchev wrote:

DK> I have sucessfully used that motherboard with FreeBSD 9 and the mps driver.
DK> The mfi driver found on the LSI site does not support this controller.

Ah that makes sense. I'm a bit reluctant to use -current on this particular
machine, so I would discuss MFCing mps driver wirh ken@

Thank you for the info! (I planned to test -current at least just booting
anyway)



DK> PS: My experiments were with the X8DTL-6F motherboard and Supermicro chassis
DK> with E16 expander. There is no reason the HBA chip in the single processor
DK> motherboard to be different.

My box is actually the same (846E1)



--
Sincerely,
D.Marck                                     [DM5020, MCK-RIPE, DM3-RIPN]
[ FreeBSD committer:                                 [hidden email] ]
------------------------------------------------------------------------
*** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [hidden email] ***
------------------------------------------------------------------------
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Francois Tigeot-2]

On Wed, Feb 16, 2011 at 07:23:26PM +0300, Dmitry Morozovsky wrote:
> On Wed, 16 Feb 2011, Daniel Kalchev wrote:
>
> DK> PS: My experiments were with the X8DTL-6F motherboard and Supermicro chassis
> DK> with E16 expander. There is no reason the HBA chip in the single processor
> DK> motherboard to be different.
>
> My box is actually the same (846E1)

Not exactly: with Supermicro, E1 means a SAS 3 Gb/s backplane and E16 is
for SAS 6Gb/s

--
Francois Tigeot
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Damien Fleuriot-2]

On 2/16/11 5:23 PM, Dmitry Morozovsky wrote:
> On Wed, 16 Feb 2011, Daniel Kalchev wrote:
>
> DK> I have sucessfully used that motherboard with FreeBSD 9 and the mps driver.
> DK> The mfi driver found on the LSI site does not support this controller.
>
> Ah that makes sense. I'm a bit reluctant to use -current on this particular
> machine, so I would discuss MFCing mps driver wirh ken@
>

Careful, even when using the mps driver, we couldn't see the *logical*
drive here, with the h200 card.

We could only use the drives by setting them to passthrough on the RAID
controller.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Dmitry Morozovsky]

On Wed, 16 Feb 2011, Francois Tigeot wrote:

FT> > DK> PS: My experiments were with the X8DTL-6F motherboard and Supermicro chassis
FT> > DK> with E16 expander. There is no reason the HBA chip in the single processor
FT> > DK> motherboard to be different.
FT> >
FT> > My box is actually the same (846E1)
FT>
FT> Not exactly: with Supermicro, E1 means a SAS 3 Gb/s backplane and E16 is
FT> for SAS 6Gb/s

Yea, you're right, it's E16 :)

--
Sincerely,
D.Marck                                     [DM5020, MCK-RIPE, DM3-RIPN]
[ FreeBSD committer:                                 [hidden email] ]
------------------------------------------------------------------------
*** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [hidden email] ***
------------------------------------------------------------------------
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Dmitry Morozovsky]

On Wed, 16 Feb 2011, Damien Fleuriot wrote:

DF> > DK> I have sucessfully used that motherboard with FreeBSD 9 and the mps driver.
DF> > DK> The mfi driver found on the LSI site does not support this controller.
DF> >
DF> > Ah that makes sense. I'm a bit reluctant to use -current on this particular
DF> > machine, so I would discuss MFCing mps driver wirh ken@
DF> >
DF>
DF> Careful, even when using the mps driver, we couldn't see the *logical*
DF> drive here, with the h200 card.
DF>
DF> We could only use the drives by setting them to passthrough on the RAID
DF> controller.

I'm perfectly ready for this: it's target for ZFS ;-P

--
Sincerely,
D.Marck                                     [DM5020, MCK-RIPE, DM3-RIPN]
[ FreeBSD committer:                                 [hidden email] ]
------------------------------------------------------------------------
*** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [hidden email] ***
------------------------------------------------------------------------
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Kenneth D. Merry]

On Wed, Feb 16, 2011 at 19:23:26 +0300, Dmitry Morozovsky wrote:
> On Wed, 16 Feb 2011, Daniel Kalchev wrote:
>
> DK> I have sucessfully used that motherboard with FreeBSD 9 and the mps driver.
> DK> The mfi driver found on the LSI site does not support this controller.
>
> Ah that makes sense. I'm a bit reluctant to use -current on this particular
> machine, so I would discuss MFCing mps driver wirh ken@

I've been planning to MFC it for a while, it has just been a time issue.

There are a few outstanding issues with the driver, but even with the
issues it would probably be better for people to be able to use it than
have nothing to use in -stable.

A few of the known issues:
 - Out of chain frames handling.  I'm working on this one now.
 - No device queue freezing on errors.  I'm working on this one.
 - No Integrated RAID handling.  This will get fixed when LSI completes
   their version of the driver.
 - Firmware upgrade code doesn't work properly.  This will get fixed when
   LSI completes their version of the driver.

For the happy path, things should work well enough I suppose.

I have attached a patch against -stable, try it out and let me know whether
it works.  If so I'll go ahead and MFC it.

Ken
--
Kenneth Merry
[hidden email]

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Dmitry Morozovsky]

On Wed, 16 Feb 2011, Kenneth D. Merry wrote:

KDM> > Ah that makes sense. I'm a bit reluctant to use -current on this particular
KDM> > machine, so I would discuss MFCing mps driver wirh ken@
KDM>
KDM> I have attached a patch against -stable, try it out and let me know whether
KDM> it works.  If so I'll go ahead and MFC it.

I'm afraid something goes wrong at your side, at least in some files in
sys/dev/mps, like

Index: sys/dev/mps/mps_ioctl.h
===================================================================
--- sys/dev/mps/mps_ioctl.h     (revision 212420)
+++ sys/dev/mps/mps_ioctl.h     (working copy)

as there is no sys/dev/mps in stable/8 kernel sources

--
Sincerely,
D.Marck                                     [DM5020, MCK-RIPE, DM3-RIPN]
[ FreeBSD committer:                                 [hidden email] ]
------------------------------------------------------------------------
*** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [hidden email] ***
------------------------------------------------------------------------
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Kenneth D. Merry]

On Wed, Feb 16, 2011 at 22:07:11 +0300, Dmitry Morozovsky wrote:

> On Wed, 16 Feb 2011, Kenneth D. Merry wrote:
>
> KDM> > Ah that makes sense. I'm a bit reluctant to use -current on this particular
> KDM> > machine, so I would discuss MFCing mps driver wirh ken@
> KDM>
> KDM> I have attached a patch against -stable, try it out and let me know whether
> KDM> it works.  If so I'll go ahead and MFC it.
>
> I'm afraid something goes wrong at your side, at least in some files in
> sys/dev/mps, like
>
> Index: sys/dev/mps/mps_ioctl.h
> ===================================================================
> --- sys/dev/mps/mps_ioctl.h     (revision 212420)
> +++ sys/dev/mps/mps_ioctl.h     (working copy)
>
> as there is no sys/dev/mps in stable/8 kernel sources

Whoops, svn diff doesn't do the right thing when there are multiple changes
merged.

I re-did the diffs manually, you should be able to do something like:

cd src
cat patch |patch -p4

Ken
--
Kenneth Merry
[hidden email]

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Damien Fleuriot-2]

On 16 Feb 2011, at 21:35, "Kenneth D. Merry" <[hidden email]> wrote:

> On Wed, Feb 16, 2011 at 22:07:11 +0300, Dmitry Morozovsky wrote:
>> On Wed, 16 Feb 2011, Kenneth D. Merry wrote:
>>
>> KDM> > Ah that makes sense. I'm a bit reluctant to use -current on this particular
>> KDM> > machine, so I would discuss MFCing mps driver wirh ken@
>> KDM>
>> KDM> I have attached a patch against -stable, try it out and let me know whether
>> KDM> it works.  If so I'll go ahead and MFC it.
>>
>> I'm afraid something goes wrong at your side, at least in some files in
>> sys/dev/mps, like
>>
>> Index: sys/dev/mps/mps_ioctl.h
>> ===================================================================
>> --- sys/dev/mps/mps_ioctl.h     (revision 212420)
>> +++ sys/dev/mps/mps_ioctl.h     (working copy)
>>
>> as there is no sys/dev/mps in stable/8 kernel sources
>
> Whoops, svn diff doesn't do the right thing when there are multiple changes
> merged.
>
> I re-did the diffs manually, you should be able to do something like:
>
> cd src
> cat patch |patch -p4
>
> Ken
> --
> Kenneth Merry
> [hidden email]
> <mps_stable.20110216.3.txt>

I am willing to test on a Dell r210 with a h200 raid card tomorrow if needed.

I destroyed the logical volume and built mps copied from -current (I opened a thread on stable@ a few weeks ago about this) but my "patch" was *nowhere* as big as this.

There's prolly tons of missed stuff... (seems to work though, just wouldn't show me the logical drive so I had to set them to passthrough)

Would that help any ?

The host being half in production now, I'd rather try stuff around only if it helps you guys...

It runs 8.2-RC3 with the sys/dev/mps/ folder from -current, a somewhat patched conf/files resembling yours, and the makefile to build mps as a module.

Let me know._______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Damien Fleuriot-2]

On 2/17/11 12:10 PM, Oliver Brandmueller wrote:

> Hi,
>
> On Thu, Feb 17, 2011 at 12:07:17PM +0100, Damien Fleuriot wrote:
>> It looks rather unhappy:
>>
>> mybsd root  /usr/ports/sysutils/smartmontools
>>                                                                  #
>> /usr/local/sbin/smartctl -a /dev/da0
>> smartctl 5.40 2010-10-16 r3189 [FreeBSD 8.2-RC3 amd64] (local build)
>> Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net
>>
>> /dev/da0: No such file or directory
>> Smartctl: please specify device type with the -d option.
>
> Thanx for the test and the quick reply! I'm really getting crazy on
> this. All the SAS controllers seems to have nifty RAID features and
> stuff... *sigh*
>
> Thanx again,
> Oliver
>

What is sad is that these controllers are becoming very mainstream now,
we're getting them more and more on Dell servers , and the fbsd project
still struggles with them (for reasons I don't know, might be LSI's
fault, might be a lack of resources or interest...)

I'm having a very hard time defending the use of fbsd for firewalls at
work, with the recent release of debian kfreebsd.

What's saving fbsd for the moment is the lack of a proper CARP
implementation on debian kfree.


Actually, cc'ing the list, I'd like to share my feelings on this.




@list:
I'm trying very hard to keep freebsd at work for our firewalls.
Recently, I installed new blade servers shipping with h700 controllers
(these attached to mfi w/o problems).

If, for some reason, we get h200 cards using the mps driver, I'll have
to install using software RAID instead.

This would be a *disaster* and extremely hard for me to justify to my boss.

The staff in the datacenters check the machines every day, making a
routing inspection to ensure no server has orange warning lights.

Typically when a hardware RAID's hard drive fails, the lights go orange
thus prompting them to warn us (in case we missed the nagios alert) for
a replacement.

If I had to use software RAID, we'd be giving up on this extra security
and, again, I'd have a very hard time justifying it.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Tom Evans-3]

On Thu, Feb 17, 2011 at 12:11 PM, Damien Fleuriot <[hidden email]> wrote:
> What is sad is that these controllers are becoming very mainstream now,
> we're getting them more and more on Dell servers , and the fbsd project
> still struggles with them (for reasons I don't know, might be LSI's
> fault, might be a lack of resources or interest...)
>
> I'm having a very hard time defending the use of fbsd for firewalls at
> work, with the recent release of debian kfreebsd.
>

Forgive my naïveté, but surely kfreebsd would have precisely the same
issues with the same controllers, since it uses FreeBSD's kernel. Am I
missing something?

Cheers

Tom
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Damien Fleuriot-2]

On 2/17/11 1:18 PM, Tom Evans wrote:

> On Thu, Feb 17, 2011 at 12:11 PM, Damien Fleuriot <[hidden email]> wrote:
>> What is sad is that these controllers are becoming very mainstream now,
>> we're getting them more and more on Dell servers , and the fbsd project
>> still struggles with them (for reasons I don't know, might be LSI's
>> fault, might be a lack of resources or interest...)
>>
>> I'm having a very hard time defending the use of fbsd for firewalls at
>> work, with the recent release of debian kfreebsd.
>>
>
> Forgive my naïveté, but surely kfreebsd would have precisely the same
> issues with the same controllers, since it uses FreeBSD's kernel. Am I
> missing something?
>
> Cheers
>
> Tom

I am not sure, but you make a good point.


Don't get me wrong though, I don't mean to offend anyone when I say the
fbsd project struggles with this driver.

There may very well be very valid reasons like lack of cooperation on
LSI's part, I wouldn't know.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Ollivier Robert]

According to Damien Fleuriot:
> I destroyed the logical volume and built mps copied from -current (I opened a thread on stable@ a few weeks ago about this) but my "patch" was *nowhere* as big as this.

Here Ken's patch:

 amd64/conf/GENERIC           |    1
 conf/files                   |    5
 dev/bwn/if_bwn.c             |    4
 dev/mps/mpi/mpi2.h           | 1121 ++++++++++++++++++
 dev/mps/mpi/mpi2_cnfg.h      | 2646 +++++++++++++++++++++++++++++++++++++++++++
 dev/mps/mpi/mpi2_hbd.h       |  114 +
 dev/mps/mpi/mpi2_history.txt |  382 ++++++
 dev/mps/mpi/mpi2_init.h      |  454 +++++++
 dev/mps/mpi/mpi2_ioc.h       | 1414 ++++++++++++++++++++++
 dev/mps/mpi/mpi2_ra.h        |   86 +
 dev/mps/mpi/mpi2_raid.h      |  302 ++++
 dev/mps/mpi/mpi2_sas.h       |  285 ++++
 dev/mps/mpi/mpi2_targ.h      |  441 +++++++
 dev/mps/mpi/mpi2_tool.h      |  391 ++++++
 dev/mps/mpi/mpi2_type.h      |   99 +
 dev/mps/mps.c                | 1821 +++++++++++++++++++++++++++++
 dev/mps/mps_ioctl.h          |  106 +
 dev/mps/mps_pci.c            |  365 +++++
 dev/mps/mps_sas.c            | 2007 ++++++++++++++++++++++++++++++++
 dev/mps/mps_table.c          |  493 ++++++++
 dev/mps/mps_table.h          |   53
 dev/mps/mps_user.c           |  944 +++++++++++++++
 dev/mps/mpsvar.h             |  390 ++++++
 dev/siba/siba_bwn.c          |    2
 dev/sis/if_sisreg.h          |    2
 mips/mips/mp_machdep.c       |    2
 modules/Makefile             |    1
 modules/mps/Makefile         |   14

There may be some patches that are not really related to mps (siba? if_sisreg?).

> It runs 8.2-RC3 with the sys/dev/mps/ folder from -current, a somewhat patched conf/files resembling yours, and the makefile to build mps as a module.

Dropping dev/mps leads to a diff close to Ken's:

 amd64/conf/GENERIC           |    1 +
 conf/files                   |    5
 dev/mps/mpi/mpi2.h           | 1121 ++++++++++++++++++
 dev/mps/mpi/mpi2_cnfg.h      | 2646 +++++++++++++++++++++++++++++++++++++++++++
 dev/mps/mpi/mpi2_hbd.h       |  114 +
 dev/mps/mpi/mpi2_history.txt |  382 ++++++
 dev/mps/mpi/mpi2_init.h      |  454 +++++++
 dev/mps/mpi/mpi2_ioc.h       | 1414 ++++++++++++++++++++++
 dev/mps/mpi/mpi2_ra.h        |   86 +
 dev/mps/mpi/mpi2_raid.h      |  302 ++++
 dev/mps/mpi/mpi2_sas.h       |  285 ++++
 dev/mps/mpi/mpi2_targ.h      |  441 +++++++
 dev/mps/mpi/mpi2_tool.h      |  391 ++++++
 dev/mps/mpi/mpi2_type.h      |   99 +
 dev/mps/mps.c                | 1821 +++++++++++++++++++++++++++++
 dev/mps/mps_ioctl.h          |  106 +
 dev/mps/mps_pci.c            |  365 +++++
 dev/mps/mps_sas.c            | 2007 ++++++++++++++++++++++++++++++++
 dev/mps/mps_table.c          |  493 ++++++++
 dev/mps/mps_table.h          |   53
 dev/mps/mps_user.c           |  944 +++++++++++++++
 dev/mps/mpsvar.h             |  390 ++++++
 modules/Makefile             |    1
 modules/mps/Makefile         |   13

--
Ollivier ROBERT -=- FreeBSD: The Power to Serve! -=- [hidden email]
In memoriam to Ondine, our 2nd child: http://ondine.keltia.net/

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Damien Fleuriot-2]

On 2/17/11 3:21 PM, Ollivier Robert wrote:
> According to Damien Fleuriot:
>> I destroyed the logical volume and built mps copied from -current (I opened a thread on stable@ a few weeks ago about this) but my "patch" was *nowhere* as big as this.
>
> Here Ken's patch:

[snip]

> There may be some patches that are not really related to mps (siba? if_sisreg?).
>
>> It runs 8.2-RC3 with the sys/dev/mps/ folder from -current, a somewhat patched conf/files resembling yours, and the makefile to build mps as a module.
>
> Dropping dev/mps leads to a diff close to Ken's:
>

[snip too]


Indeed, this looks very much more like what I have, only the conf/files
, modules makefiles, and dev/mps entries being changed.

This reassures me that I correctly copied mps from -current, and that
the inability to see the logical drive wasn't due to my incompetence ;)
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Rumen Telbizov]

Hello Damien, list:

I've been following this discussion with great interest and I hope the
result of it
will be the merge of a stable driver in STABLE branch soon.

I brought this not a long ago (
http://lists.freebsd.org/pipermail/freebsd-stable/2011-January/061237.html)
and I was hoping to have this driver in 8.x before 8.3 release.


> What is sad is that these controllers are becoming very mainstream now,
> we're getting them more and more on Dell servers , and the fbsd project
> still struggles with them (for reasons I don't know, might be LSI's
> fault, might be a lack of resources or interest...)
>

I couldn't agree more. I've been testing those (9211-8i) myself here on a
bunch of SuperMicro
servers and I'd say that I am very pleased with the performance of those
HBAs (not the mention the
price ~$230-250 CAD). On a 12 x SAS disks with a direct attached backplane
826A (6Gbit SAS2.0)
in RAID0 across all of them (md linux software raid) I was able to squeeze *
2.1GBytes/s* sequential write
(dd kind of style write). With the same setup and SATA disks I get to
1.2GByte/s. I did use 2 controllers
to accomodate ports for all the disks. I get the same speed with ZFS and
9.0-CURRENT.

On a related note I must say that I do have serious problems with the
bandwidth when I use
the SAS expander version of this backplane (826E26) (I cannot get more than
200-250 MBytes/s).
I've been investigating this problem with LSI's help and still no
conclusion. It looks like I am using
only one channel.

Anyway this was just my humble attempt to encourage the MFC of this driver.
I think the card is
pretty good. I'd like to hear other people's opinion on this HBA though.

Regards,

--
Rumen Telbizov
http://telbizov.com
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Dmitry Morozovsky]

On Wed, 16 Feb 2011, Kenneth D. Merry wrote:

KDM> > KDM> > Ah that makes sense. I'm a bit reluctant to use -current on this particular
KDM> > KDM> > machine, so I would discuss MFCing mps driver wirh ken@
KDM> > KDM>
KDM> > KDM> I have attached a patch against -stable, try it out and let me know whether
KDM> > KDM> it works.  If so I'll go ahead and MFC it.
KDM> >
KDM> > I'm afraid something goes wrong at your side, at least in some files in
KDM> > sys/dev/mps, like
KDM> >
KDM> > Index: sys/dev/mps/mps_ioctl.h
KDM> > ===================================================================
KDM> > --- sys/dev/mps/mps_ioctl.h     (revision 212420)
KDM> > +++ sys/dev/mps/mps_ioctl.h     (working copy)
KDM> >
KDM> > as there is no sys/dev/mps in stable/8 kernel sources
KDM>
KDM> Whoops, svn diff doesn't do the right thing when there are multiple changes
KDM> merged.
KDM>
KDM> I re-did the diffs manually, you should be able to do something like:
KDM>
KDM> cd src
KDM> cat patch |patch -p4

well, this sequence creates mps-related files directly in src directory, but I
managed to overcome this :)

test reboot (remotely) said that at least machine tastes mps and single
non-configured drive, but further testing is due.

Thank you very much!

--
Sincerely,
D.Marck                                     [DM5020, MCK-RIPE, DM3-RIPN]
[ FreeBSD committer:                                 [hidden email] ]
------------------------------------------------------------------------
*** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- [hidden email] ***
------------------------------------------------------------------------
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"

[Kenneth D. Merry]

On Thu, Feb 17, 2011 at 13:11:33 +0100, Damien Fleuriot wrote:

> On 2/17/11 12:10 PM, Oliver Brandmueller wrote:
> > Hi,
> >
> > On Thu, Feb 17, 2011 at 12:07:17PM +0100, Damien Fleuriot wrote:
> >> It looks rather unhappy:
> >>
> >> mybsd root  /usr/ports/sysutils/smartmontools
> >>                                                                  #
> >> /usr/local/sbin/smartctl -a /dev/da0
> >> smartctl 5.40 2010-10-16 r3189 [FreeBSD 8.2-RC3 amd64] (local build)
> >> Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net
> >>
> >> /dev/da0: No such file or directory
> >> Smartctl: please specify device type with the -d option.
> >
> > Thanx for the test and the quick reply! I'm really getting crazy on
> > this. All the SAS controllers seems to have nifty RAID features and
> > stuff... *sigh*
> >
> > Thanx again,
> > Oliver
> >
>
> What is sad is that these controllers are becoming very mainstream now,
> we're getting them more and more on Dell servers , and the fbsd project
> still struggles with them (for reasons I don't know, might be LSI's
> fault, might be a lack of resources or interest...)

LSI has actually been working on the driver, and has almost completed their
version of it.  It will go into FreeBSD in the near future.  In the mean
time I have merged the mps(4) driver into -stable.

Ken
--
Kenneth Merry
[hidden email]
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[hidden email]"