NPIV mapping problem

UNIX AIX
1

Hello all I hope someone can help me with this. I have a single LPAR which is supplied a rootvg and datavg disk set from a dual VIO server configuration. I see dual path disks for all the datavg disks but only see one path for the rootvg disks.

The client LPAR has the following:

fcs0   Available C2-T1 Virtual Fibre Channel Client Adapter
fcs1   Available C3-T1 Virtual Fibre Channel Client Adapter
fcs2   Available C4-T1 Virtual Fibre Channel Client Adapter
fcs3   Available C5-T1 Virtual Fibre Channel Client Adapter

But when lspath is run the following is returned:

Enabled hdisk4 fscsi1
Enabled hdisk5 fscsi1
Enabled hdisk6 fscsi1
Enabled hdisk2 fscsi2
Enabled hdisk0 fscsi2
Enabled hdisk4 fscsi3
Enabled hdisk5 fscsi3
Enabled hdisk6 fscsi3

As can be seen FCS0 (fscsi0) is missing. From the VIO server which is serving this adapter you have:

$ lsdev -type adapter | grep fcs
fcs0             Available   8Gb PCI Express Dual Port FC Adapter (df1000f114108a03)
fcs3             Available   8Gb PCI Express Dual Port FC Adapter (df1000f114108a03)]

When I run the map command I get the following:

$ lsmap -all -npiv
Name          Physloc                            ClntID ClntName       ClntOS
------------- ---------------------------------- ------ -------------- -------
vfchost0      U8202.E4B.06605EP-V3-C3                 2 LPAR        AIX

Status:LOGGED_IN
FC name:fcs0                    FC loc code:U78AA.001.WZSG75A-P1-C6-T1
Ports logged in:1
Flags:a<LOGGED_IN,STRIP_MERGE>
VFC client name:fcs0            VFC client DRC:U8202.E4B.06605EP-V2-C2-T1

Name          Physloc                            ClntID ClntName       ClntOS
------------- ---------------------------------- ------ -------------- -------
vfchost1      U8202.E4B.06605EP-V3-C4                 2 LPAR        AIX

Status:LOGGED_IN
FC name:fcs3                    FC loc code:U78AA.001.WZSG75A-P1-C6-T2
Ports logged in:2
Flags:a<LOGGED_IN,STRIP_MERGE>
VFC client name:fcs1            VFC client DRC:U8202.E4B.06605EP-V2-C3-T1

This is where it gets interesting on the offending vfchost it can be seen that only 1 port is logged in, whereas on the working port 2 are logged in.

I believe this is the problem but cannot find anything on a general internet trawl to answer why! I believe it is a SAN configuration problem but need evidence with which to beat up the SAN team!

I hope you can help

2

Does your client LPAR have an "Available" fscsi0 device?

hostname:/:# lsdev | grep fscsi
fscsi0         Available 02-00-02    FC SCSI I/O Controller Protocol Device
fscsi1         Available 02-01-02    FC SCSI I/O Controller Protocol Device
hostname:/:#

Also, look for the fcs0 child devices like this:

hostname:/:# lsdev -p fcs0
fcnet0 Defined   02-00-01 Fibre Channel Network Protocol Device
fscsi0 Available 02-00-02 FC SCSI I/O Controller Protocol Device
hostname:/:#
3

The simple answer is yes they are as follows:
The available adapters are:

lsdev -Cc adapter
ent4   Available       Logical Host Ethernet Port (lp-hea)
fcs0   Available C2-T1 Virtual Fibre Channel Client Adapter
fcs1   Available C3-T1 Virtual Fibre Channel Client Adapter
fcs2   Available C4-T1 Virtual Fibre Channel Client Adapter
fcs3   Available C5-T1 Virtual Fibre Channel Client Adapter
lhea0  Available       Logical Host Ethernet Adapter (l-hea)

FCS0 is from the first VIO with the partner adapter FCS2 from the second VIO
The adapter plus children are as follows:

lsdev -C | grep C2-T1
fcs0          Available C2-T1       Virtual Fibre Channel Client Adapter
fscsi0        Available C2-T1-01    FC SCSI I/O Controller Protocol Device
sfwcomm0      Available C2-T1-01-FF Fibre Channel Storage Framework Comm

The adapter from the other VIO server:

lsdev -C | grep C4-T1
fcs2          Available C4-T1       Virtual Fibre Channel Client Adapter
fscsi2        Available C4-T1-01    FC SCSI I/O Controller Protocol Device
hdisk0        Available C4-T1-01    XP MPIO Disk XP24000 (Fibre)
hdisk2        Available C4-T1-01    XP MPIO Disk XP24000 (Fibre)
sfwcomm2      Available C4-T1-01-FF Fibre Channel Storage Framework Comm
4

After a lot of work finding the connection infromation of the offending disks and using the

mkpath

and

chpath

command I finally convinced the SAN team that something was not behaving itself on the SAN switch.

A full reset was then run on the switch with all VIO servers and the client LPAR powered off. Once the switch reset had finished I rebooted the VIO servers and the client ran an

lspath

which showed all disks!

It appears that the

mkpath chpath

do not actually do anything on the fibre link between the HBA and the switch. I believe I will never find out why the ports dropped!