[md_paul]

Greetings,

New member here with a questions:

I have a VIO Server with 6 SAN luns presented to it. The 6 luns are a flash copy of a VG from another server. I need to present these 6 luns to one of the lpars on my P560 and import the volume group contained on them. However, when I create the virtual scsi luns utillizing the following comands:

mkvdev -vdev vpath1 -vadapter vhost1
mkvdev -vdev vpath2 -vadapter vhost1
mkvdev -vdev vpath3 -vadapter vhost1
mkvdev -vdev vpath4 -vadapter vhost1
mkvdev -vdev vpath5 -vadapter vhost1
mkvdev -vdev vpath6 -vadapter vhost1

and then run cfgmgr on the target lpar, the lpar has 6 new hdisks with no PVIDs. Is there a way to present these luns and maintain the PVID info needed to recreate the volume group? The PVID's are visible on the idwvio server, and I can importvg sucessfully on that server.

Any help or ideas of how do present these luns differently are appreciated!

Thanks,

- Paul

[md_paul]

Nevermind...

I spoke with IBM software support, and the answer is that it can't be done at this time.

Here is the information in the faq on IBM's VIO server site:

"Can SCSI LUNs be moved between the physical and virtual environment as is? That is, given a physical SCSI device(ie LUN), with user data on it, that resides in a SAN environment; can this device be allocated to a VIOS and then provisioned to a client partition and used by the client as is?

No, this is not supported at this time. The device cannot be used as is, virtual SCSI devices are new devices when created, and the data must be put onto them after creation. This typically would require some type of backup of the data in the physical SAN environment with a restoration of the data onto the virtual disk. "

Apparently there are a lot of users requesting this, and hopefully it will be included in a future release. It will really assist with datacenters migrating to virtual IO solutions (vs backup/restore option).

- Paul

[drax]

Anyone know if they have done anything about this? I've run into a similar problem in terms of migrating data from existing non-vio partitions into vio ones.

I guess my best bet is to tempoarily add a real fibre adapter to the target partition (more reading time!) and then present my original volume group to that. Then I should be able to extendvg/migratepv from real to virtual disk, releasing the adapter once this is done.

The lack of this functionality really does undermine one of the best features of AIX (lvm).

[lolo2clans]

Hi
Just an idea as is : did you try the fget_config -Av command to see hdisk / lun relationship and then importvg ?

[drax]

Hmm, scary no-man-page command...google...aha. I'm using HDS SAN storage and this looks like an IBM DS400 specific command.

I think I see where your headed though, and it's not going to work...

From what I've seen/read so far presenting storage via a VIO scsi device requires me to present storage as follows:

1. Present storage via SAN as per usual to the VIO partition with physical fibre adapters. Eg 32GB LUN presented and configured as hdisk2

2. Create a volume group on that storage
Eg. vvg00_rootvgs is created on hdisk2

3. Create logical volumes on that storage
Eg vlv0000 is created (32GB in size) in vvg00_rootvgs

4. Map the logical volume for use by your target virtual scsi equipped partition
Eg vlv0000 is mapped to my virtual server rs0023 and becomes hdisk0 on that server.

As you can see, I can only present logical volumes to a virtual server, not entire volume groups. Given the hdisk-ified LV appearing on the server there's no way the contents are any use even if I was crazy enough to try importing my original volume group and mapping bits of it to the server

It looks as if you have two options (after configuring free space as above):
• restore the data into the "hdisks" appearing on the virtual server
• configure a physical fibre adapter to the virtual server, present and importvg as per a "normal" AIX server. Then extendvg to your virtualized space and migratepv your data across. Reducevg and release the phsyical fibre adapter afterwards.

I'm giving option 2 a go just to see if it'll work...

Source: Redbook sg247940, section 4.4

[lolo2clans]

You can assign an entire lun to a vlpar thrue the vios, mapping the correspondant vios hdisk to the appropriate vhost .

[drax]

That sounds promising! Can you point me in the direction of some sort of explaination or command reference that explains it in detail please?

[lolo2clans]

depending on your HD bay, find on vios the relationship between the hdiskn and the lun name/nb.
you have prepare the profile of you vios and vioc with a virtual scsi on both that only communicate together.
you will have on vios a vhostx on your vscsi.
you can verify the slot nb (define on your profile) by :
lsdev -dev vhostn -field physloc this will display someting like this :
U9117.570.657B79F-V1-C8 where the C8 indicate slot 8.
then make the map by :
mkvdev -vdev hdsikn -vadapter vhostn -dev vthdiskn
you can verify the mapping by :
lsmap -vadapter vhostn

[drax]

Thanks lolo2clans that works a treat. I tried it out with a test volume group and it survived the trip with no problems!

You've answered md_paul's question and mine nicely. I wonder if this is new since the version md_paul was using or did support just drop the ball?

To sum up:
Configure Storage - Direct vios to vioc mapping of hdisks
• Login to the vios
• Determine target disk to be mapped
lspv
-present via SAN first etc
• Map an entire hdisk directly to a virtual partition (vioc)
mkvdev -vdev hdisk6 -vadapter vhost0 -dev vthdisk6
• Login to the vioc (rs0023)
• Scan for new hardware
cfgmgr
-new disk should appear with correct PVID
• Importvg

Also related:
� How to determine vioc hdisk to adapter to slot number mapping
-lsdev no longer appears to provide useful output for virtual scsi adapters and devices
• Login to vioc (client) partition (eg rs0023)
• Determine hdisks
lspv
• Determine hdisk addresses
lscfg -vpl hdisk0 | grep hdisk0
hdisk0 U9133.55A.06ED92G-V2-C4-T1-L810000000000 Virtual SCSI Disk Drive
• Determine the adapter address
lscfg -vpl vscsi0 | grep Specific
Device Specific.(YL)........U9133.55A.06ED92G-V2-C4-T1
• The "C4" in this example indicates the vioc slot number (see HMC profile for server)
• Login to the vios (server) partition (eg rs0022v0)
• Display the mappings, specific to C4
lsmap -all | grep C4
vhost0 U9133.55A.06ED92G-V1-C4 0x00000002
-This shows that vhost0 has vscsi slot 4 mapped
-From the earlier examples we now know which disk(s) are associated with this map
• Show disks mapped to vscsi
lsmap -vadapter vhost0
VTD vt0000
LUN 0x8100000000000000
Backing device vlv0000
-Results in lists of maps
-The above example shows that LV vlv0000 is mapped to vt0000 which belongs to vhost0

[lolo2clans]

The map of a complete LUN/hdisk is available since the first Version of VIOS i tryed : 1.1 (the actual version is 1.3FP8.1)
NB the new version provide a monitoring of the virtualisation based on ITM6