Yet more 10.20 to 11.00 Upgrade questions

[Donald O'Brien]

From reading this forum, I have learned to not even try the upgrade and to instead perform a cold install. So I will.

Tomorrow I have to upgrade a E55 & a D230 from 10.20 to 11.00. Using the cold-install method, I'm a therefore trying to gather and backup as much configuration information as I can before overwriting both of these systems with HP-UX 11.00. I am running the print_manifest from Ignite, and I have also written a script to gather other information (hostname, ip addr, gateway, filesystem, swapinfo, netstat -r, etc..).

I have a few questions and I hope I can beg out some answers here.

1. Is there a file that I can backup that contains the routing information? If there is, can I safely copy it back after the installation of 11.00?

2. If I have a 2 disk system, vg00 on disk1 & a second app vg on disk2, after the HP11.00 installation on vg00, will I have to restore/recreate the 2nd vg, or is there a way to just import it?

3. Which CD is OnlineJFS on? I have core, app, & support CDs.

4. I had installed HP11.00 on a D series last week and found that it was dumping to a core file in /var/adm/core. It was taking up too much diskspace and causing problems. Is there a way to keep this dump from occurring on every boot? Other E series that I have upgraded did not do this.

Thanks for any assistance.

Donald O'Brien.

[John Poff]

Hi Donald,

1. I don't think there is a single file with the route information, but it should be configured in your /etc/rc.config.d/netconf file. With a copy of that and the output from netstat -rn, you should be able to put the routing back together when you configure your LAN cards.

2. You can just vgimport in the 2nd vg. You'll have to mkdir the /dev/2nd vg directory and mknod the group file.

3. Don't know. I'll have to look.

4. Weird. What exactly was core dumping? What did 'file /var/adm/core' show you?

JP

[Pete Randall]

donald,

1) /etc/rc.config.d/netconf is the file - I would suggest that you merge it rather than just copy it back

2) Do a vgchange -a n /dev/vg01, vgexport -s -m /tmp/mapfile /dev/vg01, then copy the mapfile somewhere where you can get to it, then you can do a vgimport -s -m /tmp/mapfile /dev/vg01 (after making the /dev/vg01 directory and running mknod.

3) Not sure.

4) Not sure.

Pete

[Jeff Schussele]

Hi Donald,

1) Predefined routes will be in
/etc/rc.config.d/netconf
any routes added after will show up in the
netstat -rn
output.

2) Yes just vgexport it beforehand
vgexport -m /etc/lvmconf/vgapp.map -s -v -p /dev/vgapp
Then back that map file up, bring it back post install. Create the vg dir & group file for vgapp
mkdir /dev/vgapp
cd /dev/vgapp
mknod group c 64 0x020000
vgimport -m /etc/lvmconf/vgapp.map -s -v /dev/vgapp
Don't forget to add the entry to the /etc/fstab file to mount @ boot.

3) Should be on the Support CD I believe

4) Yes use the lvlnboot -d command to define the dump volume after load OR create a larger /var LV during initial load to hold it in case it dumps during load. We also like to link the /var/adm/crash dir to a large LV used specifically for this purpose.

HTH,
Jeff

[John Poff]

Hi again,

The Online JFS looks like it is on the app CD, on 1 of 5. There is a CD_TABLE_OF_CONTENTS file that lists all the products on all the CD's. You can mount up the CD and see where everything is in that file.

JP

[James R. Ferguson]

Hi Donald:

1. The file you want is 'etc/rc.config.d/netconf'.

In general, capture your key configuration files before you start. I like to 'diff' files in the list below, to those provided in a new release and "marry' my local needs to any enhanced features. Otherwise, many can be loaded as is.

Here's the basic list I transfer to safe-keeping for reference, reuse, and comparison against new configurations (via 'diff') during a cold installation process:

# /.forward
# /.profile
# /etc/ftpd/ftpusers
# /etc/group
# /etc/hosts
# /etc/inetd.conf
# /etc/inittab
# /etc/nsswitch
# /etc/ntp.conf
# /etc/passwd
# /etc/profile
# /etc/rc.config.d/netconf
# /etc/rc.config.d/netdaemons
# /etc/rc.config.d/nfsconf
# /etc/resolv.conf
# /etc/sendmail.cf
# /etc/services
# /etc/shells
# /etc/shutdown.allow
# /var/adm/at/*.allow|deny
# /var/adm/cron/*.allow|deny
# /var/adm/fbackupfiles/*
# /var/adm/inetd.sec
# /var/spool/cron/crontabs/*
# /sbin/init.d/

Don't forget to add your volume group 'vgexport' mapfiles too.

2. No, the whole concept of 'vgexport' and 'vgimport' is to be able to push a volume group "out" of one system and to pull it into another (or the same system to rename it, or to associate it with changed device files). This is what you will do for your non-vg00 volume groups.

Essentially, you can do this:

Use the 'mapfile' option of 'vgexport'. This allows the 'vgimport' to create logical volume devices files under the vg_name directory using the naming convention given in mapfile as opposed to the default naming convention used by the 'lvcreate'.

Before the cold-installation:

# vgexport -s -p -m /tmp/lvmconf_vgNN.map /dev/vgNN

After the cold-installation [having copied the mapfile from a tape, for instance]:

# mkdir /dev/vgNN
# mknmod /dev/vgNN/group c 64 0xNN0000
# vgimport -v -s -m /tmp/lvmconf_vgNN.map /dev/vgNN
# vgchange -a y /dev/vgNN


3. I believe that OnlineJFS is found on CD#1.


4. If you are taking memory dumps you need to have HP Support analyze "why". Open a support call to do this if this isn't a one-time occurance.

Regards!

...JRF...

[Donald O'Brien]

Thanks for all of the answers. Wish me luck.

Donald O'Brien

[Jeff Schussele]

Hi Donald,

You don't need luck.

A good plan trumps luck every single time!

Later,
Jeff

[harry d brown jr]

Though a good plan INCLUDES a way back!

get the latest version of ignite (3.7.95?) and BACK YOUR 10.20 SYSTEM UP. If you miss a configuration file, the make_tape_recovery tape can be used to extract that file!!


live free or die
harry

[Donald O'Brien]

I now have 30+ E55 & 12 D250's under my belt. They where all just vg00 only systems, but I'm glad they are done :D

I have now begun work on a dozen systems with additional volume groups. My test system here in the office is a D250 with two disks. One is vg00, the second is a vg called hekvg. I upgraded HP-UX 11.0 on vg00 and I was able to import hekvg (with the previously created mapfile copied back on the system).

I used the following commands.

# vgexport -s -p -m /tmp/lvmconf_hekvg.map /dev/hekvg

[cold install]

# mkdir /dev/hekvg
# mknod /dev/hekvg/group c 64 0x010000
# vgimport -v -s -m /tmp/lvmconf_hekvg.map /dev/hekvg
# vgchange -a y /dev/hekvg

I created the mount point and mounted up my old filesystems. Worked great!

But I have one question, I looked at the map file, and I can't see where the physical volumes are specified. How does the vgimport know which hard disks the volume group is on? Since my test box is only two disks, it is of course the only other disk which could have the hekvg, but how does vgimport know this? If I have a 3rd vg, will the system know which disks it is on?

Thanks again everyone.

Donald O'Brien

[James R. Ferguson · Accepted solution]

Hi Donald:

Since you used the '-s' option to create your mapfiles, the VGID (Volume Group ID) of the physical disk belonging to the volume group was placed in the mapfile. During 'vgimport' your physical disk were scanned for a matching ID. Those matching were "imported" and the information added to 'etc/lvmtab'.

Regards!

...JRF...

[Donald O'Brien]

Thanks for clearing that up for me. I feel much more confident now.

Donald