Upgrade 10.20 to 11i

[MarkSyder]

Hi everybody,

I have been trying to upgrade a server from 10.20 to 11i with little success.

I started by doing an fbackup so I would not lose data on the non-root disc. The tape drive appeared not to be working, so I did the fbackup to a specially created directory on the non-root disc, believing I would be able to see the disc by using vgcreate.

I then went ahead and did the upgrade only for vgcreate to tell me that if I proceed it will wipe the disc clean. Needless to say, this is not what I want! I did not proceed and now have an HP-UX 11i system which cannot see the data on the second disc.

Is there any way I can recover this?

Mark

[Chris Watkins_1]

I believe you want to do a vgimport of the disk- Not vgcreate.

Create the directory and group file for the new VG.
Note that I use vg01, but you can use any non-used vg name.

# mkdir /dev/vg01
# mknod /dev/vg01/group c 64 0x010000

Import that disk as vg01 (Substitute correct instances for "X")

# vgimport vg01 /dev/dsk/cXtXdX


(Or, if you happen to have a mapfile)

# vgimport -m /path/to/mapfile vg01


You should now be able to create your mountpoint(s)
and view the old data you'd saved off to that disk.

[Bill Hassell]

No problem. SAM to the rescue! (did I say that?) Start SAM and select Disks and File Systems -> Volume Groups. Now select Actions from the top menu and choose Import. SAM will scan all the disks, find the other volume group and give you the opportunity to change the default name (vg01). If you like the naqme, select OK, you'll get a message about alternate links which you can ignore and the volume group will now be reconnected and activated. And if you used the standard names for the original lvols (ie, lvol1 lvol2 lvol3...) then you can simply mount everything.

Even if you had oddball lvol names, you can rename the device files without any special considerations: mv /dev/vg01/lvol1 /dev/vg01/myvol1

Now there is thye expert way to do this using vgexport and vgimport but you should have done a vgexport prior to the upgrade.

[Norman_21]

My friend,

Let me put my $0.2 cents over here.

Install the HD in another server and mount it, then copy the backup over to a tape.

In the future, what should be done is :

#vgexport -p -v -s -m /etc/vg01.map /dev/vg01
#vgexport -p -v -s -m /etc/vg02.map /dev/vg02

create a recovery tape:
#make_tape_recovery -x inc_entire=vg00 -I -v

When you restore just restore the tape and then:
#vgimport -p -v -s -m /etc/vg01.map /dev/vg01
#vgimport -p -v -s -m /etc/vg02.map /dev/vg02

Hope this help

[Tim D Fulford]

Hi

Some great advice above... Just a smidgen more info though. You can import your disk(s) into a new volume group BUT without the map file the logical volumes will simply be thedefault names, labled /dev/vgxx/lvol1 .. lvol2 .. lvol3 .. etc. This is not really a problem as you can mount them on "dummy" mount points to determine what each one is.

Good luck, you made the right decision in not doing a vgcreate & you are close to being where you inteded.

Regards

Tim

[MarkSyder]

OK, we're making progress.

The server can now see the volume group and 8 logical volumes, but the LVs are all shown as unused. How do I mount them on a dummy mountpoint?

All help so far appreciated, further help will be equally appreciated.

Mark

[MarkSyder]

So near yet so far!

I've found out how to make a dummy mount point by typing "dummy mount point" into google and following the link.

I can now see the files on the second disc and, more importantly, I can see the information from the first disc that I backed up to a safe place on Friday. I have copied the original /etc/fstab into place and paged it. And that's where I encounter my latest problem.

The logical volumes are not numbered logically (no pun intended). In the fstab file the volumes for the first disc are numbered 1 - 8 and those for the second disc are 10 - 17. The upgraded system is seeing both discs as having volumes 1 - 8. How do I persuade it to see the volumes on the second disc as 10 - 17?

While I wait for an answer, I will amend the fstab file (having made a safe copy first) to see if changing the logical volume numbers for the second disc will work.

Mark

[Paula J Frazer-Campbell]

HI

If your search this forum you will see over and over again - Do not upgrade but do a fresh install.

There are many threads on how to prepare for this but bacically.

1. Fully back up you server TWICE.
2. Make an Ingite tape of your server TWICE.

Then go for a fresh install.

Paula

[Bill Hassell · Accepted solution]

It sounds like you have a simple server with no extra LAN cards which is one of the biggest obstacles in upgrading from 10.anything to 11.anything, so things sound like they are working OK. The good news is that your second volume group is just fine, imported successfully and now you just need to make the lvols match fstab. Two easy choices:

1. Change /etc/fstab to use the new numbering in the second volume group. Just vi the file and change the source devicefile names for each mountpoint. Once complete, mount each one using just the destination mountpoint name and then check the contents. Once everything is mounted successfully, you can reboot and everything will be back to normal.

2. You can rename the device files for the lvols. Assuming you have a /dev/vg01 directory for the second disk, just use mv as in:

mv /dev/vg00/lvol1 /dev/vg00/lvol10
mv /dev/vg00/rlvol1 /dev/vg00/rlvol10

Make sure you rename *BOTH* device files. The 'r' volumes are the raw devicefiles and are required when performing specialized operations such as fsck and newfs.

The adbvice to cold-install may still be required. You'll need to checkout your uograded system fairly carefully. The 10.20-to-11.anything process has a number of failure points, some of which are obscure. Be sure you run the cleanup command using the -i option (see the man page for cleanup). ALSO: your 11i upgrade is not complete until you add the patch bundles for HWE and Gold QPK. And is you do not have programs in /usr/sbin called cstm and xstm, be sure to load the DIAGNOSTICS from your quarterly SupportPlus CDROM. If yuo don't have a subscription to software updates, you can download the 2 HWE and Gold QPK packages from the ITRC at http://www1.itrc.hp.com/service/patch/locateBundles.do but unfortunately, the DIAGNOSTICS bundles have been removed. (yet another reason to get a software subscription service)

[MarkSyder]

Thanks Bill - I think I'm nearly there.

Just a little clarification - I have misled people by using the word "upgrade" - I know that upgrading doesn't work, but I described it as an upgrade because I am changing the OS on a live server. Apologies for the confusion.

Tape backup failed - I'm going to investigate why when I've got the server working properly (and allocated points to everyone who has helped me). I did an ignite backup over the network, but when I tried to restore from it it failed with "bad magic" and "bad lif" - that's why I'm in the position of having to get the upgrade (oops, there's that word again!) working - there's no going back!

Thanks again for the suggestions.

[MarkSyder]

Thanks for the suggestions everyone.

I've now got the second disc fully on line and only need to get the networking sorted - I've moved over to the networking forum - any networking experts, please join me there!

Mark