Strange issue about IBM SDD on HP-UX (vpath devices).

[mpua]

Hi!

We have some boxes in my company with the following specifications:

HP-UX 11.23
IBMsdd Multipathing software (datapath)

The SDD script starts up at runlevel 1:

/sbin/rc1.d/K150CONFserver
/sbin/rc1.d/S200CONFserver

Loading the VPATH module and translating disk devices (c#t#d#s) into vpath devices:

Dev#: 0 Device Name: vpath1 Type: 2145 Policy: Optimized
Serial: 6005076801AB824E5000000000000089
================================================================================
Path# Adapter H/W Path Hard Disk State Mode Select Error
0 6/0/9/1/0/4/0 c4t0d1 OPEN NORMAL 1183685 0
1 6/0/9/1/0/4/0 c2t0d1 OPEN NORMAL 0 0

Dev#: 1 Device Name: vpath2 Type: 2145 Policy: Optimized
Serial: 6005076801AB824E500000000000016B
================================================================================
Path# Adapter H/W Path Hard Disk State Mode Select Error
0 6/0/9/1/0/4/0 c4t0d0 OPEN NORMAL 19802 0
1 6/0/9/1/0/4/0 c2t0d0 OPEN NORMAL 0 0


In the boxes, we have a ServiceGuard clustering environment so that I've disabled AUTO_VG_ACTIVATION in /etc/lvmrc for obvious reasons (I dont' want ServiceGuard managed volume groups to be autostarted on all nodes). I've specified the VG's I want to be enabled by hand:
This is an excerpt:

AUTO_VG_ACTIVATE=0

#
# The variable RESYNC controls the order in which
# Volume Groups are resyncronized. Allowed values
# are:
# "PARALLEL" - resync all VGs at once.
# "SERIAL" - resync VGs one at a time.
#
# SERIAL will take longer but will have less of an
# impact on overall I/O performance.
#

RESYNC="SERIAL"


#
# Add customized volume group activation here.
# A function is available that will synchronize all
# volume groups in a list in parallel. It is
# called parallel_vg_sync.
#
# This routine is only executed if AUTO_VG_ACTIVATE
# equals 0.
#

custom_vg_activation()
{
# e.g. /sbin/vgchange -a y -s
# parallel_vg_sync "/dev/vg00 /dev/vg01"
# parallel_vg_sync "/dev/vg02 /dev/vg03"
/sbin/vgchange -a y /dev/vg01
/sbin/vgchange -a y /dev/vg02

return 0
}


When I reboot the box, the following error prompts on the console screen:

/sbin/bcheckrc:
Checking for LVM volume groups and Activating (if any exist)
vgchange: Warning: Couldn't attach to the volume group physical volume "/dev/dsk/vpath2":
The path of the physical volume refers to a device that does not
exist, or is not configured into the kernel.
vgchange: Warning: couldn't query physical volume "/dev/dsk/vpath2":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query all of the physical volumes.
vgchange: Couldn't activate volume group "/dev/vg01":
Quorum not present, or some physical volume(s) are missing.

vgchange: Warning: Couldn't attach to the volume group physical volume "/dev/dsk/vpath1":
The path of the physical volume refers to a device that does not
exist, or is not configured into the kernel.
vgchange: Warning: couldn't query physical volume "/dev/dsk/vpath1":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query all of the physical volumes.
vgchange: Couldn't activate volume group "/dev/vg02":
Quorum not present, or some physical volume(s) are missing.


When the server is started, u can see that vg01 and vg02 are not activated but, since the VPATH module is load and SDD started, I can enable them by hand with "vgchange -a y" without problem....

This means that for some reason I dont know... the disk to vpath translation is happening after VG activation, leading to the impossibility to enable those VGs...

I need guru advice here!

[Shibin_2]

Any clue at rc.log ?

[mpua]

Its not particularly useful rc.log in this case. It simply says that the service is started:

Created VPAR device files

Starting vpath loading and vpath_config
Output from "/sbin/rc1.d/S200CONFserver start":
----------------------------
NOTE: The automatic 'backup' configuration currently contains the
configuration that was in use before the last reboot of this
system. It has not been updated since boot.
NOTE: No change to 'vpath' was needed.
Module State Cause Notes
vpath loaded explicit loadable, unloadable

Mount file systems
Output from "/sbin/rc1.d/S250localmount start":

[mpua]

Does anyone has faced this problem with SDD Multipathing?

[TTr]

A few things come to mind

SDD has some kernel patch prerequisites. Ensure they are installed.

I used sdd long time ago and if I remember correctly it is a kernel driver that binds the multiple paths into one virtual device. So your problem may not be SDD related.

SDD did not support all levels of serviceguard. Check on the IBM site to find out if your OS/SDD/SD versions are compatible. This is the result of the following. When the v-devices get created in each server in the cluster, the v-devices that have the same name must also contain the same physical paths.

[mpua]

Ok, I'll check that.

Thanks.