Re: "exportvg -p -m" for AIX

• Previous message: Drew Myers: "Re: "exportvg -p -m" for AIX"
• Maybe in reply to: Drew Myers: ""exportvg -p -m" for AIX"
• Next in thread: Bill Verzal: "Re: "exportvg -p -m" for AIX"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date:         Fri, 27 Feb 2004 08:52:36 -0600
To: aix-l@Princeton.EDU

We are doing pretty much the same thing using ESS Shark Disk Subsystems and Flashcopy (ESS/Copy Services functionality). Every night we perform a "flash" of our source drives to the destination drives. Once the replication to the destination drives is complete we simply do an importvg on what would be your LPAR #2. As long as your disks or powerpath devices have PVIDs on them on LPAR #2 everything should work fine. We then mount all of the filesystems on LPAR #2 and backup the "copy" of our production data to tape. Once the backup is complete we unmount the filesystems and export the volume groups on LPAR #2 and everything is in place for the "flash" to take place the following night.

Hope that helps.

Jason de la Fuente

>>> myersd@VANGUARDCAR.COM 02/27/04 07:18AM >>>
All,

Thanks very much for all of your helpful comments and suggestions.

After thinking about this more, and trying the suggestions presented here, as well as my own, I think I've narrowed the problem a bit. It also seems I didn't present enough information to make analyzing the issue very easy. Here's the complete "skinny".

2 LPARs on a p671 running AIX 5.2 with EMC Symmetrix attached LUNs via fiber.
I use EMC's Control Center applications (timefinder et al), to manipulate the LUNs.

I'm going to simplify the actual architecture for illustrative purposes.

+--------+ +-----------------+
| LPAR 1 | | Symmetrix |
+--------+ +-----------------+
| lun001 |-------->| dev001 | bcv001 |
| lun002 |-------->| dev002 | bcv002 |
| lun003 |-------->| dev003 | bcv003 |
+--------+ +--------+--------+
                              | | |
+--------+ | | |
| LPAR 2 | | | |
+--------+ | | |
| lun001 |--------------------+ | |
| lun002 |-----------------------+ |
| lun003 |--------------------------+
+--------+

One this *hopefully* conveys, is that I have 3 LUNs visible to LPAR1, from the Symmetrix. These are "standard" devices, all the same size. They are not visible to LPAR2. BCV1-3 are "BCV" devices, annotated as such at the Symmetrix.

My goal, is to establish a mirror, via the Symmetrix, between dev001-3 and bcv001-3, then split that mirror. This part works fine. I then would like to be able to access the data on bcv001-3 on LPAR2 for backup and development purposes. Most of the data is raw, but there are a few mounted filesystems, which seems to be where I run into trouble. Based on something Bill Verzal mentioned, I did notice that the VGIDs on the LUNs are identical on LPAR1 and LPAR2.

What would be the appropriate method to make the data on bcv001-3 (and therefore lun001-3 on LPAR2) accessible on LPAR2? I assume the data is present, I just need to inform the ODM the VGs, LVs, and filesystems are there.

Sorry to bring this up again, but I hope this explanation proves more instructive than my last.

Thanks again,

Drew

• Previous message: Drew Myers: "Re: "exportvg -p -m" for AIX"
• Maybe in reply to: Drew Myers: ""exportvg -p -m" for AIX"
• Next in thread: Bill Verzal: "Re: "exportvg -p -m" for AIX"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]