druddle

Anyone got any experience with using these 2 products together, creating VxVM volumes on the Powerpath pseudo devices ?

Dave

DrEvil

Dave,

Personally, I have limited knowledge due to limited use and IBM ESS creeping in..

But if you post the query here or email me at work, I'll try and get an answer for you off my colleague.

Rgds, Alex

druddle

Basically i have /dev/dsk/emcpower0a - 46a configured with Powerpath.

I am configuring VxVM host based mirroring, and need to initialise the LUNS presented to the host.

I am using VxVM 3.5.

Do i run the vxdiskadd on the emcpower device or the c2t0d0s2 device ?

Cheers

Dave

krankyd

I'm sure you use the powerpath device.

Use vxassist and see which device options it gives you when you ask it to add a disk.

DMP and powerpath do co-exist (with the correct versions) OK.

druddle

Yeah i am sure too. But when i use vxdiskadm to initialise the disks it only sees the c4t2d12 type devices. There is something in the Powerpath manual about getting VxVM to see the pseudo devices but it doesnt say what they are

Dave

krankyd

Hi,

the only thing I could find was this below - sorry if the formatting is poo:

In VERITAS Volume Manager (VxVM) 3.2 and 3.5, if EMC PowerPath is being used with emcpower names and those devices are not excluded from Dynamic Multipathing (DMP), then various WARNING and ERROR messages will be displayed when VxVM commands are run.
Exact Error Message:

Various non-fatal WARNING and ERROR messages from Volume Manager commands when emcpower devices are used

Solution:

Examples of commands and the error messages:

1. Typical messages from vxdg:

# vxdg init optdg disk3=emcpower3c disk4=emcpower4c
vxvm:vxconfigd: WARNING: vold_dm_as_da: Error (22) from dmp_getpri_subpath
for device 0x368001a
vxvm:vxconfigd: WARNING: vold_dm_as_da: Error (22) from dmp_getpri_subpath
for device 0x3680022


2. Typical messages from the vxdisk list:

# vxdisk list emcpower1c
Device: emcpower1c
devicetag: emcpower1c
type: simple
hostid: nyptpsn01a.eq.gs.com
disk: name=disk1 id=1007878664.1275.nyptpsn01a.eq.gs.com
group: name=userdg id=1007879118.1278.nyptpsn01a.eq.gs.com
info: privoffset=1
flags: online ready private autoimport imported
pubpaths: block=/dev/vx/dmp/emcpower1c char=/dev/vx/rdmp/emcpower1c
version: 2.2
iosize: min=512 (bytes) max=256 (blocks)
public: slice=0 offset=1025 len=17674495
private: slice=0 offset=1 len=1024
update: time=1007879401 seqno=0.6
headers: 0 248
configs: count=1 len=727
logs: count=1 len=110
Defined regions:
config priv 000017-000247[000231]: copy=01 offset=000000 enabled
config priv 000249-000744[000496]: copy=01 offset=000231 enabled
log priv 000745-000854[000110]: copy=01 offset=000000 enabled
vxvm:vxdisk: ERROR: emcpower1c not a valid dmpnode disk access name.


3. Typical messages from vxassist:

# vxassist -g optdg make optvol 10g
vxvm:vxassist: ERROR: emcpower3c not a valid dmpnode disk access name.
vxvm:vxassist: ERROR: emcpower4c not a valid dmpnode disk access name.


All of these messages are non-fatal and the commands will complete the operations that have been requested.

These messages occur because the emcpower devices are not seen as DMP devices to VERITAS Volume Manager.

DrEvil

Dave
Did you install powerpath before or after VxVM?

The only suggestion my colleague came up with was that you need to install & config EMCpowerpath prior to installing VxVM.

I'll check SunSolve...

Alex

DrEvil

SRDB ID Synopsis Date
27684 Veritas Volume Manager 3.X - Unable to Access/Initialize disks under Solaris and EMC PowerPath 28 Mar 2002

Problem Description Top

For Volume Manager 3.X versions only.

The disks had previously been working, but now they are inaccessible
while booted to the Solaris OS, and I/O errors are being logged.
When booted off the Solaris cdrom, all the disks are readable and writable. Problem Solution Top

Customer is running EMC PowerPath 2.0. There is a bug associated
with this software which makes all Non-EMC disks inaccessible unless they have
a slice 0 (which disappears if you initialize the disk for SEVM and therefore
renders the disk unusable). EMC also has a patch which can be loaded to
resolve this PowerPath bug.

Keywords: VxVM, Volume Manager, disks, disk initialization, configure, configuration, ECM disks

druddle

Alex - installed Powerpath first then VxVM. Did it that way after being bitten by Powerpaths predecessor on Clariion, ATF

After speaking to a colleague in Germany yesterday this is the answer i got:

You should use the Solaris native device. PP puts an "invisible" driver layer between what Solaris reports and the sd driver, so that if you talk to the c0t1d0 device then you are actually talking to the Powerpath device. This was confirmed when i ran a dd to the c0t4d0 device and i saw activity on 4 of the ports on the FC switch, indicating that the IO was travelling down all 4 ports that PP reported were configured for the emcpower0a device.

You can use the emcpower0a device, but EMC advise against it. You can only use the emcpower0a device as a simple disk, with no partitions.

Dave

krankyd

hmmm..

That sounds logical - like using c0t0d0s0 devices with veritas as opposed to the DMP devices.





[Edited by krankyd - 3/6/2003 10:42:02 AM]