Operating System - HP-UX
1754014 Members
7360 Online
108811 Solutions
New Discussion

Can a volume group created on HP-UX be imported on a Red Hat Linux system?

 

Can a volume group created on HP-UX be imported on a Red Hat Linux system?

Does anyone know if this is possible? I have an HP-UX 11iv3 box that we are using to export a database. We then NFS mount that export filesytem over to a Red Hat box to import the data. However this takes a very long time to pull all of that data over the network. It's prohibitively long in fact in that it takes almost a day. So I want to see if it's possible to take the SAN space presented to the HP-UX box and present it directly to the Linux box and then do a vgimport to see all of that data.

 

Thanks!

1 REPLY 1
Matti_Kurkela
Honored Contributor

Re: Can a volume group created on HP-UX be imported on a Red Hat Linux system?

Linux LVM is not compatible with HP-UX LVM, so a vgimport will not be possible.

 

However, you could theoretically create a filesystem directly on top of the SAN LUN, bypassing LVM completely.

You would need to use a filesystem type that is supported for both HP-UX and Linux.

On HP-UX, the filesystem choices are very limited, and only VxFS is supported for large filesystems.

 

Linux has a FreeVxFS driver which apparently can support VxFS layout versions 2-4 in read-only mode.

Looks like VxFS layout 4 is the absolutely oldest VxFS layout that can be supported by HP-UX 11.31.

So if all the stars align perfectly, it *might* work. But that layout version has a maximum filesystem size limit of 2 TB. Will it be enough for your purposes?

 

If you are willing to pay, you can get a commercial VxVM and VxFS implementation for Linux from Symantec. If you also use VxVM on HP-UX, this might be the least painful way to do this.

 

Remember that you still won't be able to access the filesystem simultaneously from HP-UX and Linux: doing that would require coordination between the systems. When one system updates some data the other has in its disk cache, the

other node must be made aware of the change. Otherwise there is a risk that the other node will combine stale data from its cache with newer data from disk, corrupting your data.

MK