How to avoid the probe-scsi-all command hanging the system when there are multiple hosts on the same Fibre Channel Loop

Asset ID:	1-71-1021453.1
Update Date:	2010-01-25
Keywords:

Solution Type Technical Instruction Sure

Solution 1021453.1 : How to avoid the probe-scsi-all command hanging the system when there are multiple hosts on the same Fibre Channel Loop

Related Items


Sun Storage 3510 FC Array

Related Categories


GCS>Sun Microsystems>Storage - Disk>Modular Disk - 3xxx Arrays

PreviouslyPublishedAs
271929

Description
At the OBP prompt, the probe-scsi-all command may cause host(s) to hang,
when there are multiple hosts connected to the same fibre channel loop.

When we configure hosts to boot from the StorageTek[tm] 3510 FC array  that are configured on
the same fibre channel loop, if a probe-scsi-all command is issued at the OBP prompt on one host, other hosts might not be able to boot from OBP due to CR: 4589231's effect.
CR 4589231 probe-scsi-all ; boot -v kills LILP map and topology discovery process
Also see related CR: 6904697 Synopsis: probe-scsi-all should cause the hosts hang  RFE.
If other hosts are running, it might cause those hosts to hang with 'Loop OFFLINE' message.
The following is the message outputted from a running host when issued
probe-scsi-all on the other server.
Sep 23 17:25:59 xxxxxx lw8: [ID 555425 kern.notice] Main, up 53 days
00:51:16, Memory 8,755,544
Sep 23 21:25:59 xxxxxx lw8: [ID 944324 kern.notice] Main, up 53 days
04:51:16, Memory 8,591,008
Sep 24 00:57:59 xxxxxx qlc: [ID 630585 kern.info] NOTICE: Qlogic
qlc(2): Loop OFFLINE
Sep 24 05:35:38 xxxxxx genunix: [ID 540533 kern.notice] ^MSunOS
Release 5.10 Version Generic_118833-24 64-bit
Sep 24 05:35:38 xxxxxx genunix: [ID 172907 kern.notice] Copyright
1983-2006 Sun Microsystems, Inc. All rights reserved.
We tested in the following environment.
[env]
V480 x2 + SE3510
V480-a:PCI#0 ---- SE3510:Controller#A CH0
V480-a:PCI#1 ---- SE3510:Controller#B CH1
V480-b:PCI#0 ---- SE3510:Controller#A CH1
V480-b:PCI#1 ---- SE3510:Controller#B CH0
- HBA： P/N(375-3363), FW(1.17), Jumper(1-2)
- SE3510：Controller P/N(371-0532), FW(4.21F)
--------------------
test 1 (reproduced)
--------------------
(1) v480a poweron
(2) v480b poweron
(3) v480a probe-scsi-all
(4) v480b probe-scsi-all
(5) v480a boot <-- boot fail
(6) v480b boot <-- boot fail
--------------------
test 2（reproduced, though changed v480a v480b boot order）
--------------------
(1) v480b poweron
(2) v480a poweron
(3) v480b probe-scsi-all
(4) v480a probe-scsi-all
(5) v480b boot <-- boot fail
(6) v480a boot <-- boot fail
--------------------
test 3（not reproduced if one node boot earlier）
--------------------
(1) v480a poweron
(2) v480a probe-scsi-all
(3) v480a boot
(4) v480b poweron
(5) v480b probe-scsi-all
(6) v480b boot
--------------------
test 4（not reproduced if issue reset-all after probe-scsi）
--------------------
(1) v480a poweron
(2) v480b poweron
(3) v480a probe-scsi-all
(4) v480a reset-all
(5) v480b probe-scsi-all
(6) v480b reset-all
(7) v480a boot
(8) v480b boot
--------------------
Test log snip
--------------------
- power on v480a and v480b
- doing probe-scsi-all on v480a
===================
{2} ok probe-scsi-all
/pci@9,600000/SUNW,qlc@2
LiD HA LUN --- Port WWN --- ----- Disk description -----
0 0 0 2100000c50ac5725 SEAGATE ST336605FSUN36G 0638
1 1 0 2100000c50696a0c SEAGATE ST336607FSUN36G 0307
/pci@8,600000/SUNW,qlc@2,1
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
Link not ready - Loss of Sync
/pci@8,600000/SUNW,qlc@2
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
*************************** Local Loop Devices ***************************
Adapter LoopId - 0
Device AL-PA a5 loopId 2a Hard-addr 2a Port WWN 226000c0ffa01f53
Lun 0 SUN StorEdge 3510 413C
Device AL-PA a3 loopId 2b Hard-addr 2b Port WWN 226000c0ffb01f53
Lun 0 SUN StorEdge 3510 413C
/pci@8,600000/SUNW,qlc@1,1
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
Link not ready - Loss of Sync
/pci@8,600000/SUNW,qlc@1
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
*************************** Local Loop Devices ***************************
Adapter LoopId - 0
Device AL-PA a7 loopId 28 Hard-addr 28 Port WWN 216000c0ff801f53
Lun 0 SUN StorEdge 3510 413C
Device AL-PA a6 loopId 29 Hard-addr 29 Port WWN 216000c0ff901f53
Lun 0 SUN StorEdge 3510 413C
===================
Issue probe-scsi-all on v480b.
===================
{2} ok probe-scsi-all
/pci@9,600000/SUNW,qlc@2
LiD HA LUN --- Port WWN --- ----- Disk description -----
0 0 0 21000004cf98a7bd SEAGATE ST373405FSUN72G 0638
1 1 0 21000004cf989c39 SEAGATE ST373405FSUN72G 0638
/pci@8,600000/SUNW,qlc@2,1
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
Link not ready - Loss of Sync
/pci@8,600000/SUNW,qlc@2
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
Link not ready - Waiting for AL_PA
/pci@8,600000/SUNW,qlc@1,1
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
Link not ready - Loss of Sync
/pci@8,600000/SUNW,qlc@1
QLA2342 Host Adapter fcode version 1.17 08/09/07
Firmware version 3.03.23
Link not ready - Waiting for AL_PA
===================

Steps to Follow
The following steps are the suggested workaround.

If all systems on the same fibre channel loop are at the OBP prompt: issue
a 'reset-all' after performing the "probe-scsi-all" command.
If any systems on the shared fibre channel loop are up and running,
do not issue a 'probe-scsi-all' without first shutting down those machines.

Product
Sun StorageTek 3510 FC Array

probe-scsi-all, shared loop, hung, se3510, minnor, 3510, 3510 Array, OBP
Product_uuid
58553d0e-11f4-11d7-9b05-ad24fcfd42fa

Attachments

This solution has no attachment