topic Re: Automated BCV replication in Operating System - HP-UX

Automated BCV replication

Mike Smith_2 — Mon, 26 Feb 2001 19:25:20 GMT

I'm trying to put together a script to automate EMC BCV replication, and I can make it work the first time, but none after that.

The first run through, the VG for the BCV volumes gets created, mknod creates the group file, vgchgid does its thing, vgimport imports the volumes, vgchange makes it active, vgcfgbackup back up the configuration, fsck checks the filesystem, and it's mounted.

Now, I have a replicated dataset. I want to replicate it again for the next day. So, I umount the filesystem, re-issue the establish command, split it when done, only now I can't seem to get LVM to see them again.

Out of the original process that created everything, what parts of it do I need to tun to mount it again?

TIA

Mike

Re: Automated BCV replication

James A. Donovan — Mon, 26 Feb 2001 19:48:18 GMT

You not only need to unmount your BCV volume group(s), you need to do something like:

# vgchange -a n /dev/vgbcv01
# vgexport /dev/vgbcv01

then you re-establish your STD-BCV pairs. If you don't then the information under /dev/vgbcv01 (or whatever you call your BCV volume group) will be out of sync with reality.

The next day you start the process all over again, by splitting the BCV's, and rebuilding your volume group.

Re: Automated BCV replication

Mike Smith_2 — Mon, 26 Feb 2001 20:30:24 GMT

Wow, the vgexport command you list effectively deletes it. So, I have re-create the whole thing each time? Hmm.. And they said it was easier...

Re: Automated BCV replication

James A. Donovan — Mon, 26 Feb 2001 21:03:05 GMT

Yes, you have to do it all over again every time. Remeber, when you synchronize your BCV's, it's synchronizing the WHOLE disk, not just the part where your files reside.

This is why you need to issue the vgchgid command, because when you split off the BCV volume that device has the same Volume Group ID as the STD device it had been mirrored with.

Re: Automated BCV replication

Mike Smith_2 — Mon, 26 Feb 2001 21:12:34 GMT

Right. I understand that part of it. I was just having trouble with the imported nature of the volumes. You'd think that once the physical paths were part of a given volume group, and vgchgid made them viewable, that you should just be able to fsck and mount them.

I guess not...

Thanks.