Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

SDDPCM query : pcmpath query device <n>

Status
Not open for further replies.

raylin

MIS
Jun 7, 2001
177
0
0
CH
What are meaning the stars in the path column in the query :
pcmpath query device 12

DEV#: 12 DEVICE NAME: hdisk12 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F8800000000000007
===========================================================================
Path# Adapter/Path Name State Mode Select Errors
0 fscsi0/path0 OPEN NORMAL 26484 2
1 fscsi0/path1 OPEN NORMAL 26593 2
2* fscsi0/path2 OPEN NORMAL 54 2
3* fscsi0/path3 OPEN NORMAL 54 2
4 fscsi1/path4 OPEN NORMAL 12 0
5 fscsi1/path5 OPEN NORMAL 12 0
6* fscsi1/path6 OPEN NORMAL 8 0
7* fscsi1/path7 OPEN NORMAL 8 0
 
3.2 New Features

3096 Mark non-preferred paths in 'pcmapth query device', 'pcmpath query essmap', and 'pcmpath query portmap' commmands.

see the readme file that came with the SDDPCM driver:
Code:
3.3  Feature/defect Details

     3096	(PCR 2886)Mark non-preferred paths in 'pcmpath query device', 
		'pcmpath query essmap', and 'pcmpath query portmap' commands.

		There are three pcmpath commands that offer this feature:
			pcmpath query device
			pcmpath query essmap
			pcmpath query portmap

		The following is a sample output of this new option in 'pcmpath
		query device'.

		> pcmpath query device 5

		DEV#: 0  DEVICE NAME: hdisk5 TYPE: 1750500 ALGORITHM:Load Balance
		SERIAL: 13AAAKA1200
		====================================================================
		Path#    Adapter/Hard Disk    State     Mode     Select    Errors
		    0*       fscsi1/hdisd5     OPEN   NORMAL          0         0
		    1        fscsi0/hdisk5     OPEN   NORMAL    5571118         0
		    2*       fscsi0/hdisk5     OPEN   NORMAL          0         0
		    3        fscsi1/hdisk5     OPEN   NORMAL    5668419         0
		
		In 'pcmpath query essmap' output, non-preferred paths are marked with 
		an asterisk under the column 'P'.

		Disk    Path  P  Location   adapter    LUN SN       Type        Size  LSS  Vol  Rank  C/A   S ...
		------  ----  - ----------  ------   ------ -----  -----------  ----  ---- ---  ----- ----  - ...
		hdisk5  path0 * 30-60-01[FC] fscsi1   13AAAKA1200 IBM 1750-500  1.1   18    0   0000   01   Y ...
		hdisk5  path1   30-60-01[FC] fscsi0   13AAAKA1200 IBM 1750-500  1.1   18    0   0000   01   Y ...
		hdisk5  path2 * 20-60-01[FC] fscsi0   13AAAKA1200 IBM 1750-500  1.1   18    0   0000   01   Y ...
		hdisk5  path3   20-60-01[FC] fscsi1   13AAAKA1200 IBM 1750-500  1.1   18    0   0000   01   Y ...


		In 'pcmpath query portmap' output, non-preferred paths are designated by
		lower case letters 'y', 'o' and 'n'. 

				  BAY-1(B1)             BAY-2(B2)             BAY-3(B3)              BAY-4(B4)		
		ESSID  DISK   H1   H2   H3   H4     H1   H2   H3   H4     H1   H2   H3   H4      H1   H2   H3   H4
			      ABCD ABCD ABCD ABCD   ABCD ABCD ABCD ABCD   ABCD ABCD ABCD  ABCD   ABCD ABCD ABCD ABCD
			          BAY-5(B5)             BAY-6(B6)             BAY-7(B7)              BAY-8(B8)
			      H1   H2   H3   H4     H1   H2   H3   H4     H1   H2   H3   H4      H1   H2   H3   H4
			      ABCD ABCD ABCD ABCD   ABCD ABCD ABCD ABCD   ABCD ABCD ABCD  ABCD   ABCD ABCD ABCD ABCD
	       13AAAKA hdisk5 O--- ---- ---- ----   o--- ---- ---- ----   ---- ---- ----  ----   ---- ---- ---- ----
	       13AAAKA hdisk6 Y--- ---- ---- ----   y--- ---- ---- ----   ---- ---- ----  ----   ---- ---- ---- ----

		Y  =  online/open               y = (alternate path) online/open
		O  =  online/closed             o = (alternate path) online/closed
		N  =  offline                   n = (alternate path) offline
		-  =  path not configured
		PD =  path down

HTH,

p5wizard
 
I migrated Lun's on ESS to Lun's on DS4700+SVC with
vg mirroring. I had never this situation before.
I think it has a performance impact.
Can a reboot clear the things because I migrated the filesystems without application downtime.

 
Well you have 8 paths for each LUN, it is normal that some paths are automatically disused, as long as some paths down each FC adapter are used.

You still have 2 paths down fscsi0 and 2 paths down fscsi1 which are unmarked, so they are in use - you're OK on that. But the load-balancing numbers are a bit strange. Do you have the same difference in the "Select" column for each hdisk or just this one? Maybe you added the paths via fcs1/fscsi1 just recently?

What version of AIX, MPIO and SDDPCM are you running?


HTH,

p5wizard
 
AIX 5300-04-00
SDDPCM devices.sddpcm.53.rte 2.1.1.0
devices.fcp.disk.ibm.mpio.rte 1.0.0.5

The string * on the path column is alternatively on
path 0,1,4,5 on dev n
path 2,3,6,7 on dev n+1

DEV#: 2 DEVICE NAME: hdisk2 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F880000000000000A
==========================================================================
Path# Adapter/Path Name State Mode Select Errors
0* fscsi0/path0 OPEN NORMAL 30 0
1* fscsi0/path1 OPEN NORMAL 30 0
2 fscsi0/path2 OPEN NORMAL 1114 0
3 fscsi0/path3 OPEN NORMAL 1063 0
4* fscsi1/path4 OPEN NORMAL 30 0
5* fscsi1/path5 OPEN NORMAL 30 0
6 fscsi1/path6 OPEN NORMAL 1106 0
7 fscsi1/path7 OPEN NORMAL 1073 0

DEV#: 3 DEVICE NAME: hdisk3 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F880000000000000B
==========================================================================
Path# Adapter/Path Name State Mode Select Errors
0 fscsi0/path0 OPEN NORMAL 145 0
1 fscsi0/path1 OPEN NORMAL 145 0
2* fscsi0/path2 OPEN NORMAL 30 0
3* fscsi0/path3 OPEN NORMAL 30 0
4 fscsi1/path4 OPEN NORMAL 152 0
5 fscsi1/path5 OPEN NORMAL 159 0
6* fscsi1/path6 OPEN NORMAL 30 0
7* fscsi1/path7 OPEN NORMAL 30 0

DEV#: 4 DEVICE NAME: hdisk4 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F880000000000000C
==========================================================================
Path# Adapter/Path Name State Mode Select Errors
0* fscsi0/path0 OPEN NORMAL 30 0
1* fscsi0/path1 OPEN NORMAL 30 0
2 fscsi0/path2 OPEN NORMAL 153 0
3 fscsi0/path3 OPEN NORMAL 146 0
4* fscsi1/path4 OPEN NORMAL 30 0
5* fscsi1/path5 OPEN NORMAL 30 0
6 fscsi1/path6 OPEN NORMAL 150 0
7 fscsi1/path7 OPEN NORMAL 151 0

DEV#: 12 DEVICE NAME: hdisk12 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F8800000000000007
===========================================================================
Path# Adapter/Path Name State Mode Select Errors
0 fscsi0/path0 OPEN NORMAL 69466 4
1 fscsi0/path1 OPEN NORMAL 69471 4
2* fscsi0/path2 OPEN NORMAL 62 3
3* fscsi0/path3 OPEN NORMAL 61 3
4 fscsi1/path4 OPEN NORMAL 43149 1
5 fscsi1/path5 OPEN NORMAL 43035 1
6* fscsi1/path6 OPEN NORMAL 15 1
7* fscsi1/path7 OPEN NORMAL 15 1

DEV#: 13 DEVICE NAME: hdisk13 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F8800000000000008
===========================================================================
Path# Adapter/Path Name State Mode Select Errors
0* fscsi0/path0 OPEN NORMAL 37 1
1* fscsi0/path1 OPEN NORMAL 37 1
2 fscsi0/path2 OPEN NORMAL 32990 1
3 fscsi0/path3 OPEN NORMAL 32967 1
4* fscsi1/path4 OPEN NORMAL 16 1
5* fscsi1/path5 OPEN NORMAL 15 1
6 fscsi1/path6 OPEN NORMAL 2820 1
7 fscsi1/path7 OPEN NORMAL 2747 1

DEV#: 14 DEVICE NAME: hdisk14 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 600507680191016F8800000000000009
===========================================================================
Path# Adapter/Path Name State Mode Select Errors
0 fscsi0/path0 OPEN NORMAL 1852770 12
1 fscsi0/path1 OPEN NORMAL 1853338 7
2* fscsi0/path2 OPEN NORMAL 43 1
3* fscsi0/path3 OPEN NORMAL 43 1
4 fscsi1/path4 OPEN NORMAL 1556614 6
5 fscsi1/path5 OPEN NORMAL 1556785 6
6* fscsi1/path6 OPEN NORMAL 15 1
7* fscsi1/path7 OPEN NORMAL 15 1



 
You are load-balancing the IO load fairly evenly between both FC adapters. There is really no reason to have more than 2 paths per FC adapter active, because the FC adapter would do just as many IOs, except that on the AIX side, you'd see the IOs distributed among the 4 paths running out of on FC adapter, whereas now 2 of the 4 paths are favoured.

Perhaps the LUNs have a "semi-dedicated" FC adapter on the DS4700 side, and if the LUNs are not evenly distributed among the DS4700 FC adapters, that may be a bottleneck.

Sorry, I don't have enough (near zero) knowledge about DS4000 range.


HTH,

p5wizard
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top