Multi path on Solaris 8, EVA 5000 vcs 4.044 and FCA2257p

[Sven Bo Rønn_1]

Hi
I'm trying to connect an Sun Fire V280R to our EVA 5000 Array. VCS is 4.044. Solaris is ver 8.
HBA's are FCA2257P.
I cant get Suns MPxIO solution to work in this configuration. I've installed SUN SFS 4.4.11. HBA driver ver 4.20.
I have presented a lun to the host, and it sees it four times - one for each path. But when I use cfgadm I see none of the san luns.
I have scrolled through a lot of different information on this, and now i'm not sure that this is a supported solution.
Does anybody know if my setup is supposed to work??
In some documents I can read that Suns MPxIO only works with Suns original HBA's, but in other documents FCA2257 is also mentioned.
If it's not supported, how do I then get multipath support?
Do I have to change my HBA's or do I have to switch to Veritas DMP og SecurePath?
On Qlogic site I have found fcode for qlogic 2310F HBA, which is newer than the fcode on my HP FCA2257P. Is it an good idea to upgrade my HP hba's with qlogic fcode? We are not booting from SAN.
Any help appreciated.
Regards
Sven Roenn, Spar Nord

[Michael Steele_2]

Dear Sven:

From what I know about qLogic I think its installation will help your administration in all cases, so I think you're on the right track by installing it.

[Sven Bo Rønn_1]

Michael,
I just received an answer from an HP rep here in Denmark, and he also adviced us to upgrade the FCA2257 with qlogic fcode, so that will be our next move.
Sven

[Craig_83]

Sven,

What does your /kernel/drv/scsi_vhci.conf look like?

Also, what were the cfgadm parameters?

if you do a "luxadm -e port" is the status "CONNECTED"?

Craig

[Jonathan Harris_3]

We had a batch lot of 2257P's that we bought when Solaris was only supported with SecurePath on EVA 3000.

However, we've successfully been using these cards with EVA 4000 / 6000, Solaris 10 and MPxIO.

I realise that doesn't explicitly answer your question, but at least demonstrates that it does work.

[Sven Bo Rønn_1]

Craig,

Q1: What does your /kernel/drv/scsi_vhci.conf look like?

name="scsi_vhci" class="root";
mpxio-disable="no";
load-balance="round-robin";
auto-failback="disable";
device-type-scsi-options-list =
"HP HSV101", "symmetric-option",
"COMPAQ HSV111", "symmetric-option",
"HP HSV2", "symmetric-option",
"HP OPEN", "symmetric-option";
symmetric-option = 0x1000000;

Q2: Also, what were the cfgadm parameters?
cfgadm -al -o show_FCP_dev

Q3: if you do a "luxadm -e port" is the status "CONNECTED"?
yes

Sven

[Craig_83]

Hi Sven,

Here is what I have in scsi_vhci.conf:

name="scsi_vhci" class="root";
mpxio-disable="no";
load-balance="round-robin"
auto-failback="disable";
device-type-scsi-options-list =
"COMPAQ HSV111","symmetric-option",
"HP HSV210", "symmetric-option2";
symmetric-option = 0x1000000;
symmetric-option2 = 0x1000000;

I have many servers at Solaris 8 and 9 that utilize the FCA2257 without any problems. What kernel patch level do you have? My Solaris 8 systems are at 117350-24

Have you tried using luxadm with the following commands?

Issue a "luxadm probe". You should see something like this:
Found Fibre Channel device(s):
Node WWN:50001fe15000e2e0 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40000C26200020000009B0000d0s2
Node WWN:50001fe15000e2e0 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40000C26200020000007D0000d0s2
Node WWN:50001fe15000e2e0 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40000C2620002000000950000d0s2
Node WWN:50001fe15000e2e0 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40000C2620002000000820000d0s2
Node WWN:50001fe15000e2e0 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40000C2620002000000980000d0s2
Node WWN:50001fe15006b230 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40010522E0000800001BD0000d0s2
Node WWN:50001fe15006b230 Device Type:Disk device
Logical Path:/dev/rdsk/c5t600508B40010522E0000800001C40000d0s2

From the above output, using the logical path, issue a "luxadm display -v" path. Taking the first device,
Luxadm display -v :/dev/rdsk/c5t600508B40000C26200020000009B0000d0s2
DEVICE PROPERTIES for disk: /dev/rdsk/c5t600508B40000C26200020000009B0000d0s2
Vendor: COMPAQ
Product ID: HSV111 (C)COMPAQ
Revision: 4004
Serial Num: Unavailable
Unformatted capacity: 153600.000 MBytes
Read Cache: Enabled
Minimum prefetch: 0x0
Maximum prefetch: 0x0
Device Type: Disk device
Path(s):

/dev/rdsk/c5t600508B40000C26200020000009B0000d0s2
/devices/scsi_vhci/ssd@g600508b40000c26200020000009b0000:c,raw
Controller /devices/pci@1d,700000/QLGC,qla@1/fp@0,0
Device Address 50001fe15000e2e9,5
Host controller port WWN 210000e08b08cc82
Class primary
State ONLINE
Controller /devices/pci@1d,700000/QLGC,qla@1/fp@0,0
Device Address 50001fe15000e2ed,5
Host controller port WWN 210000e08b08cc82
Class primary
State ONLINE
Controller /devices/pci@1d,700000/QLGC,qla@2/fp@0,0
Device Address 50001fe15000e2e8,5
Host controller port WWN 210000e08b129b96
Class primary
State ONLINE
Controller /devices/pci@1d,700000/QLGC,qla@2/fp@0,0
Device Address 50001fe15000e2ec,5
Host controller port WWN 210000e08b129b96
Class primary
State ONLINE

Craig

[Sven Bo Rønn_1]

Hi Craig
We are still stuck here :-(
I hope you can help us with some hints. Solaris 8 kernel patch level is 117350-44.

cfgadm -al og luxadm probe commands only presents us for the internal 2 fc attached disks.

pi693pa30:~$ > luxadm probe
No Network Array enclosures found in /dev/es

Found Fibre Channel device(s):
Node WWN:20000004cf094d60 Device Type:Disk device
Logical Path:/dev/rdsk/c1t0d0s2
Node WWN:2000002037e9977c Device Type:Disk device
Logical Path:/dev/rdsk/c1t1d0s2
pi693pa30:~$ > cfgadm -al
Ap_Id Type Receptacle Occupant Condition
c0 scsi-bus connected configured unknown
c0::dsk/c0t6d0 CD-ROM connected configured unknown
c1 fc connected configured failed
c1::21000004cf094d60 disk connected configured failing
c1::2100002037e9977c disk connected configured failing
c2 scsi-bus connected unconfigured unknown

Right now we have only one FCA2257 in the server. But the results of the commands are not diferent from when we had 2 HBA's mounted.

The Fcode on the HBA has been upgraded to 2.00.09

In format we can see the lun. At the moment there's 1 lun presented to the HBA.

format
AVAILABLE DISK SELECTIONS:
0. c1t0d0
/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w21000004cf094d60,0
1. c1t1d0
/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w2100002037e9977c,0
2. c3t1d2
/pci@8,700000/QLGC,qla@3/sd@1,2
3. c3t2d2
/pci@8,700000/QLGC,qla@3/sd@2,2

This is the lines from logfile messages:
Jan 4 06:40:09 pi693pa30 genunix: [ID 936769 kern.info] ohci0 is /pci@8,700000/
usb@5,3
Jan 4 06:40:12 pi693pa30 qla2300: [ID 818750 kern.notice] QLogic Fibre Channel
Driver v5.01 Instance: 1
Jan 4 06:40:12 pi693pa30 qla2300: [ID 572349 kern.notice] hba1: QLogic Fibre Ch
annel Host Adapter fcode version 2.00.09
Jan 4 06:40:12 pi693pa30 qla2300: [ID 389430 kern.info] hba1: powered on
Jan 4 06:40:14 pi693pa30 qla2300: [ID 647389 kern.info] hba1: Fibre Channel Loo
p is Down (8013)
Jan 4 06:40:15 pi693pa30 qla2300: [ID 726912 kern.info] hba1: 2Gb Fibre Channel
Loop is Up (8014)
Jan 4 06:40:18 pi693pa30 qla2300: [ID 419863 kern.notice] hba1: Firmware v3.3.2
1 (ipx)

The lun is formatted, mounted and in use.

Regards,
Sven

[Peter Mattei · Accepted solution]

The FCA2257P is not supported with MPxIO but with DMP on Solaris and VCS 4.xxx.

The following HBAs are supported for VCS4.x, Solaris 8 and MPxIO:

Emulex: LP10000, LP10000DP, LP11000, LP11002

Qlogic: QLA2310F, QLA2340, QLA2342, QLA2360, QLA2362

SUN: SG-XPCI1FC-QF2, SG-XPCI2FC-QF2, SG-XPCI1FC-QL2, SG-XPCI2FC-QF2-Z, SG-XPCI1FC-QF4, SG-XPCI2FC-QF4, SG-XPCI1FC-EM2, SG-XPCI1FC-EM4-Z, SG-XPCI2FC-EM4-Z

Cheers
Peter

[Sven Bo Rønn_1]

Hi
That explains why the FCA2257 we moved to an Solaris 9 box are working fine.
I think that we will upgrade our Solaris 8 - had to be done anyway - now we have a good reason ;-)
regards
Sven

[Peter Mattei]

Does it work with MPxIO on Solaris9??
In our support matrices this is not supported either!(VCS4.x, Solaris8 or 9 or 10, FCA2257P, NO MPxIO) Again only DMP is supported!

Cheers
Peter

[Sven Bo Rønn_1]

Peter,
I must admit, that at the moment we're not running MPxIO. Right now we only have one HBA in each box. We started with two FCA2257 in one Solaris 8 box, but couldn't get it up running (I think we did some mistakes, mixing drivers and software from different versions). Then we moved one of the HBA to another SUN V280R running Solaris 9, and followed the instructions in the document *HP StorageWorks MPxIO for Sun Solaris - application notes*. It now reacts according to the notes and as expected.
We think that adding HBA nr 2 will not give any problems, but have not tried yet.
Regards
Sven

[Neitzner]

Hi, using MPxIO implies to use the SUN qlc driver instead of qla2300. Even though qlc should be able to handle 2257P HBAs all guides I found advise to use the qla2300 driver. So I would be very interested in Jonathan Harris experiences, as he seems to run lots of 2257P HBAs with qlc and MPxIO.

Regards, Marc