U320 drives reverting to async. mode, then disappearing

I have:

X4200

Solaris 10 11/06 fully patched

Sun (LSI) PCI-X Single U320 SCSI HBA (SGXPCI1SCSILM320-Z), LSI MPTBIOS 5.07.03

2 x Seagate ST3146707LC U320 SCA 10K drives

Supermicro CSE-M35S U320/U160 SCSI backplane (with QLogic GEM318 SAF_TE)

LSI BIOS identifies Seagate drives (target 0, 1), GEM318 (target 6) and adapter (target 7). I have created an entry in /kernel/drv/mpt.conf to drop mpt1 to U160, as QLogic GEM318 backplane only supports U160:

name="mpt" parent="/pci@1,0/pci1022,7450@2"

unit-address="1"

scsi-options=0x1ff8;

The backplane is terminated. When Solaris boots, I see:

Apr 23 13:29:02 myhostTarget 0 reducing sync. transfer rate

Apr 23 13:29:02 myhost scsi: WARNING: /pci@1,0/pci1022,7450@2/pci1000,5110@1 (mpt1):

Apr 23 13:29:02 myhostTarget 0 reverting to async. mode

Apr 23 13:29:02 myhost scsi: WARNING: /pci@1,0/pci1022,7450@2/pci1000,5110@1 (mpt1):

Apr 23 13:29:02 myhostTarget 1 reducing sync. transfer rate

Apr 23 13:29:02 myhost scsi: WARNING: /pci@1,0/pci1022,7450@2/pci1000,5110@1 (mpt1):

Apr 23 13:29:02 myhostTarget 1 reverting to async. mode

The drives do not appear when I run format. They are not listed in prtpicl -v or prtconf -v. However, if I insert an old Sun 9.0GB drive (Fujitsu MAB3091SC) as target 2, Solaris discovers all drives and they run at FAST40 speed:

# prtpicl -v | egrep "NAME=|sync"

| NAME=mpt0 |

| NAME= usb_mid0 Power |

| NAME= hid0 Power |

| NAME= scsa2usb0 Power |

| NAME=spindle-motor |

| NAME= scsa2usb1 Power |

| NAME=spindle-motor |

| NAME= hubd0 Power |

| NAME=spindle-motor |

| NAME=mpt1 |

:target0-sync-speed40000

:target1-sync-speed40000

:target2-sync-speed40000

:target6-sync-speed5000

| NAME=spindle-motor |

I have swapped the cable, the HBA, the disks and the backplane with same results. Any ideas how can I run this at U160 without the redundant Sun 9.0GB drive?

Thanks.

[2069 byte] By [rweatherleya] at [2007-11-27 1:57:04]
# 1
Did you ever determine your issue? I have a similar problem on different equipment. I have a blade 2000 that is intermittently dropping the hard drives. Same as you, format reports no disks. Reboot and 50% of the time it will find them but they will not be online long.--Tony
TonyBraya at 2007-7-12 1:32:12 > top of Java-index,Storage Forums,Storage General Discussion...
# 2

No resolution yet. I have logged a support request. I have also ordered another Sun Single Channel U320 PCI-X HBA and a Sun 68 pin/VHDCI SCSI cable and will test again with that. I have also found that the X4200 must be hard reset (unplug power cables at rear) in order to properly reset the bus.

The JBOD has two separate channels/backplanes. The odd thing is that the second bus is connected to a Sun Dual Channel PCI U320 HBA in a old E220R running patched Solaris 8 SPARC. There are four drives in the bay: 2 x U320 Seagates and 2 x Sun 9.0GB Fujitsu (FAST40MB/s). The E220R has the recent mpt patch. All SCSI devices clock down to 40MB/s and the bus behaves properly. No bus errors after 2 weeks.

This beats me. If I get a resolution, I will post it. Good luck.

Richard

rweatherleya at 2007-7-12 1:32:12 > top of Java-index,Storage Forums,Storage General Discussion...
# 3
The solution for me was to get rid of the third party U320 SCSI cables and use Sun cables (part# X3832A-Z). Not a single error, both channels running at U320 for over a week. HTH.Richard
rweatherleya at 2007-7-12 1:32:13 > top of Java-index,Storage Forums,Storage General Discussion...