PV link,clariion and HP-UX 10.20

[Mauro Gatti]

Hi all, I have a HP-UX 10.20 server connected to a CX700.
Because under HP-UX 10.20 powerpath isn't supported I have to use pvlink to "survive" to a lun trespass.
In my VG however I have a strange behaviour that I'd like to discuss with you:

These are two paths that I have to link in my VG:

# ioscan -fnC disk
disk 70 10/8/1/0.2.22.0.0.0.1 sdisk CLAIMED DEVICE DGC CX700WDR5
/dev/dsk/c8t0d1 /dev/rdsk/c8t0d1
disk 71 10/8/1/0.2.26.0.0.0.1 sdisk CLAIMED DEVICE DGC CX700WDR5
/dev/dsk/c9t0d1 /dev/rdsk/c9t0d1

It seems my server sees correctly LUN by this paths:

# diskinfo /dev/rdsk/c8t0d1
SCSI describe of /dev/rdsk/c8t0d1:
vendor: DGC
product id: CX700WDR5
type: direct access
size: 83886080 Kbytes
bytes per sector: 512
root@rimilx9 [/]
# diskinfo /dev/rdsk/c9t0d1
SCSI describe of /dev/rdsk/c9t0d1:
vendor: DGC
product id: CX700WDR5
type: direct access
size: 83886080 Kbytes
bytes per sector: 512

but when I ask LVM about my VG I get:

# vgdisplay -v vg_prdtn
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c9t0d1":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
--- Volume groups ---
VG Name /dev/vg_prdtn
VG Write Access read/write
VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 10239
VGDA 2
PE Size (Mbytes) 8
Total PE 10238
Alloc PE 10238
Free PE 0
Total PVG 0

vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c9t0d1":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
--- Logical volumes ---
LV Name /dev/vg_prdtn/lv_prdtn_01
LV Status available/syncd
LV Size (Mbytes) 81904
Current LE 10238
Allocated PE 10238
Used PV 1


--- Physical volumes ---
PV Name /dev/dsk/c8t0d1
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c9t0d1":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
PV Status available
Total PE 10238
Free PE 0
Autoswitch On


Now current path in use is c8.
If my server can sees disk both from c8 and from c9 path (as diskinfo says) why does LVM seems to sees disk only from "in use" path?

Thank You

RGDS

Mauro

[Carlos Roberto Schimidt]

Hi,

You can show output from /etc/lvmpvg and "strings /etc/lvmtab" about these disks?

You run this command for add another path?:

#vgextend /dev/vg_prdtn /dev/dsk/c9t0d1

[Mauro Gatti]

lvmtab contains all paths inlcuded in VG
There is no lvmpvg

[Patrice Le Guyader]

Hello,

I wonder if the disk which is under the LUN in the array hasn't been changed.
The disk should be claimed and if it's the same size diskinfo is the same. Unfortunately the LVM structure is lost.

Just an idea.

Regards
Pat

[Victor BERRIDGE]

Hi Mauro,
A (very) long time since I touched a clariion sub system so I may not be much of a help...
If I understand what you are trying to achieve, you want to acccess a same PV by 2 separate controllers, am I right?
On is the current the other is the alternate in case of failure.
If it were the case then you should see at vgcreate time that the second disk beeing extended would be given as alternate,( I will look at a 10.20 K class here if I can find old logs of creation...)
SO I wonder if you are not in a misconfigured case "clariion" side where the disk you are seeing in c9 is truly another LUN

Just thoughts
.. going to see what I can find

All the best
Victor

[Victor BERRIDGE]

Mauro,
What kind of box is it?
Could you list your interfaces?
e.g.
ioscan -funC ext_bus

[Mauro Gatti]

I think Patrice ideas can be correct.
In past Clariion was well confugured then somethin came up.
Now we reconfigured CX box with EMC support in correct way but unluckly we found out this problem.
My server is a K calss

# ioscan -funC ext_bus
Class I H/W Path Driver S/W State H/W Type Description
===================================================================================
ext_bus 0 10/0 c720 CLAIMED INTERFACE GSC built-in Fast/Wide SCSI Interface
ext_bus 8 10/8/1/0.2.22.0.0 fcparray CLAIMED INTERFACE FCP Array Interface
ext_bus 6 10/8/1/0.2.22.255.0 fcpdev CLAIMED INTERFACE FCP Device Interface
ext_bus 9 10/8/1/0.2.26.0.0 fcparray CLAIMED INTERFACE FCP Array Interface
ext_bus 7 10/8/1/0.2.26.255.0 fcpdev CLAIMED INTERFACE FCP Device Interface
ext_bus 3 10/12/0 CentIf CLAIMED INTERFACE Built-in Parallel Interface
/dev/c3t0d0_lp
ext_bus 2 10/12/5 c720 CLAIMED INTERFACE Built-in SCSI


What about reducing (I hope -f vgreduce option will work fine)the VG and then extending it againg on alternate path?

[Tiziano Contorno _]

Hi Mauro, have you tried:

pvchange -s /dev/dsk/c9t0d1

?

This changes to primary an alternate path.

Regards.

[Victor BERRIDGE]

Hi Mauro,
The reason I think is a misconfiguration LUN side is to my experience with HDS subsystems, disks can fail and you change them without any connected box noticing anything, otherwise there is no point inveting in such equipment.
So changing disks should be transparent OS side (it shouldnt even know you had a failed disk...)

Now you said you did some intervention with EMC support I believe even more in misconfiguration...
Can you get the EMC support to check again the LUN affectation layout?

Ah about your idea of vgreduce - I would give a try!

Good luck