Does anybody have experience with a MD1420, mpr(4)

[Lee Brown]

On Thu, Nov 16, 2017 at 10:43 AM, Lee Brown <leeb at ratnaling.org> wrote:

>
>
> On Thu, Nov 16, 2017 at 6:28 AM, Kenneth D. Merry <ken at freebsd.org> wrote:
>
>> On Wed, Nov 15, 2017 at 12:24:47 -0800, Lee Brown wrote:
>> > Hi,
>> >
>> > This is my for foray into DAS, so please be gentle :)  Apologies if the
>> > cross-post is inappropriate, let me know.
>> >
>> > I'm looking to build a SAS array with a Dell MD1420.  It will be
>> attached
>> > via a SAS 9300-414E HBA in a Dell R330.
>> >
>> > I've verified that the server / SAS card should be fine, but just
>> wanted to
>> > know if anybody has set this up before, or if there's any obvious reason
>> > why it wouldn't work.  Dell can be a bit weird sometimes.
>> >
>> > Any information is most welcome.
>>
>> In theory it should be fine.  I haven't used a Dell enclosure recently,
>> but
>> from posts here, it seems that they like to turn on Type 2 Protection
>> Information on the drives.  (Assuming you buy the drives from Dell.)
>>
>> If they do that, you can either reformat, or see the patch here for the
>> mpr(4) driver to disable using protection information:
>>
>> https://docs.freebsd.org/cgi/getmsg.cgi?fetch=308820+0+archi
>> ve/2017/freebsd-scsi/20171022.freebsd-scsi
>>
>> I haven't gotten around to changing things so that we detect the PI in the
>> da(4) driver and use 32 byte CDBs when type 2 protection is enabled.  (So
>> if someone else has time and wants to do it, feel free.  It will be
>> somewhat messy because we only allocate enough space for 16 byte CDBs
>> inside
>> the CCB, so you'll need to allocate the CDB separately...)
>>
>> If you use ZFS, you'll have checksum protection on your data anyway, so PI
>> would just be additional protection.
>>
>> I would also suggest that you update to the latest firmware (currently
>> Phase 15) for the SAS controller.  You can do that with LSI's sas3flash or
>> the FreeBSD mprutil(8) utility.
>>
>> Ken
>> --
>> Kenneth Merry
>> ken at FreeBSD.ORG
>>
> Thanks Ken,
>
> I'm going to purchase the system.  I had no idea about PI, that's a nice
> feature for sure.
>
> This will probably just be iSCSI and NFS4 over a 10Gb NIC (Xen cluster),
> so having PI detection is certainly of interest to me.  I'll figure out
> what I can do to contribute (code/testing) when things have settled.
>
> Assuming all is well I'll get the 1420 added to the hardware compatibility
> list.
>
> Much appreciated -- lee
>

Following up, I have the MD1420 installed and attached.  The kernel patch
works as expected, errors with the flag set to zero, no errors when set to
1.

First problem I am having is saving firmware from the card to local storage:
root at svr-nas-rl:~ # mprutil flash save firmware
mprutil: Fail to save firmware

Second problem is performance, 20MB/s sequential write to a single spindle,
200MB/s read.  I was expecting at least 100MB/s for the write, maybe my
expectations were wrong?

root at svr-nas-rl:~ # cp /dev/zero /dev/da0
dT: 1.058s  w: 1.000s
 L(q)  ops/s    r/s   kBps   ms/r    w/s   kBps   ms/w   %busy Name
    1    158      0      0    0.0    158  20206    6.3   99.8| da0

root at svr-nas-rl:~ # cp /dev/da0 /dev/null
dT: 1.001s  w: 1.000s
 L(q)  ops/s    r/s   kBps   ms/r    w/s   kBps   ms/w   %busy Name
    1   1587   1587 203171    0.6      0      0    0.0   99.4| da0

root at svr-nas-rl:~ # mprutil show adapter
mpr0 Adapter:
       Board Name: SAS9300-4i4e
   Board Assembly: H3-25515-00G
        Chip Name: LSISAS3008
    Chip Revision: ALL
    BIOS Revision: 6.00.00.00
Firmware Revision: 5.00.00.00
  Integrated RAID: no

root at svr-nas-rl:~ # geom disk list da0
Geom name: da0
Providers:
1. Name: da0
   Mediasize: 300000000000 (279G)
   Sectorsize: 512
   Mode: r0w0e0
   descr: SEAGATE ST300MM0078
   lunid: 5000c500a0f3f977
   ident: W0K1BE3R
   rotationrate: 10000
   fwsectors: 63
   fwheads: 255