PowerPath on Linux different outputs.

[rsmith_1]

We are working on a common script to determine if both the PATHS are present (in PowerPath for HP-UX, AIX, Solaris and Linux), While the ouput of powermt display dev=all shows the the disk# and h/w path on all versions of Unix, it shows only the instance of the hba driver on Linux. Anyone noticed this so far? It is different only on Linux with PowerPath. Here is the ouput from a Linux host

# powermt display dev=all
Pseudo name=emcpowera
Hitachi ID=14642
Logical device ID=HITACHI R50139320021
state=alive; policy=NoRedirect; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
5 qla2xxx sdb 8A active alive 0 0
3 qla2xxx sde 1A active alive 0 0

Pseudo name=emcpowerb
Hitachi ID=14642
Logical device ID=HITACHI R50139320022
state=alive; policy=NoRedirect; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
5 qla2xxx sdc 8A active alive 0 0
3 qla2xxx sdf 1A active alive 0 0


Could someone throw more light on this?

Thanks

Roger.

[Matti_Kurkela]

Linux does not *have* a standard and all-encompassing notion of a hardware path, so the powermt command obviously cannot display it.

For example, the disk/LUN "emcpowera" in your listing is known as "HITACHI R50139320021" by the storage system (whose ID is 14642). It is connected through two paths: /dev/sdb and /dev/sde. Both paths are currently alive and active. The paths go through two different qla2xxx HBAs.

I don't currently have a Linux system with PowerPath available to me, so I cannot test my theory; but I think the numbers "5" and "3" might refer to the Linux storage controller numbers. If that's correct, you'll find more information about each adapter in the appropriate directories of the /sys virtual filesystem. For example, the HBA 5 would have its FibreChannel-specific information at /sys/class/fc_host/host5/ and SCSI-compatible information at /sys/class/scsi_host/host5/.

If you want to know the PCI slot IDs of each HBA, the command "ls -l /sys/class/scsi_host" might be helpful.

If your Linux distribution includes the "lsscsi" tool, install it: it is useful in identifying any SCSI-like storage devices (i.e. pretty much all of them these days...), not just strictly SCSI devices.

Was this the sort of light you were looking for?

MK

[rsmith_1]

MK

Thanks for your reply. Could you please expand on your statement "Linux does not *have* a standard and all-encompassing notion of a hardware path"??

Here is what we are trying to accomplish to see if we are able to see both the PATHS alive ..Here are the outputs from AIX and Solaris also..



Pseudo name=hdiskpower0
Symmetrix ID=000190103657
Logical device ID=36F4
state=alive; policy=SymmOpt; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
1 fscsi2 hdisk24 FA 13dB active alive 0 0
0 fscsi0 hdisk6 FA 4dA active alive 0 0

Pseudo name=hdiskpower1
Symmetrix ID=000190103657
Logical device ID=36F5
state=alive; policy=SymmOpt; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
1 fscsi2 hdisk25 FA 13dB active alive 0 0
0 fscsi0 hdisk7 FA 4dA active alive 0 0



Pseudo name=emcpower28a
Symmetrix ID=000190104754
Logical device ID=00B3
state=alive; policy=SymmOpt; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
3088 pci@25,700000/SUNW,emlxs@0/fp@0,0 c10t50060482D5312C96d1s0 FA 7bA active alive 0 0
3075 pci@5,700000/SUNW,emlxs@0/fp@0,0 c3t50060482D5312C96d1s0 FA 7bA active alive 0 0
3079 pci@13,700000/SUNW,emlxs@0/fp@0,0 c5t50060482D5312C96d1s0 FA 7bA active alive 0 0

Pseudo name=emcpower27a
Symmetrix ID=000190104754
Logical device ID=00B4
state=alive; policy=SymmOpt; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
3088 pci@25,700000/SUNW,emlxs@0/fp@0,0 c10t50060482D5312C96d2s0 FA 7bA active alive 0 0
3075 pci@5,700000/SUNW,emlxs@0/fp@0,0 c3t50060482D5312C96d2s0 FA 7bA active alive 0 0
3079 pci@13,700000/SUNW,emlxs@0/fp@0,0 c5t50060482D5312C96d2s0 FA 7bA active alive 0 0

Thanks

Roger.