Change IP addresses and wehn reboot box freezes.

[Matthew Pegge_1]

I have 2 servers an old (live) box and a new server that will replace the current live one. They were both happily on the LAN together with different IP addresses and hostnames. After some data migration, I then changed the files /etc/hosts and /etc/rc.config.d/netconf, so that they swapped IP's and hostnames. I then rebooted both boxes simultaneously. One box, the newer faster one, reboots and I can log on fine. Until the older slower box boots up. The session then freezes, I can however continue to work on the old box. I know this sounds like a duplicate IP or hostanem but I have checked this 100 times and there is no duplication in either /etc/hosts and /etc/rc.config.d/netconf.

This is 1 of about 20 other succeful migrations we have completed so I am a bit stuck! Any ideas why this is happening? Both boxes are running HP-UX 11.00.

[Pete Randall]

Matthew,

I always use "/sbin/set_parms ip_address" to change ip addresses lest I forget some obscure place that will cause me pain later on.


Pete

[RolandH]

Could it be a problem in your DNS configuration ?

Perhaps on your old box you have a configuartion in /etc/nsswitch.conf that is
hosts: dns [NOTFOUND=continue] files
and the dns shows still always the old IP address for the old box. Than change /etc/nsswitch.conf to
hosts: files [NOTFOUND=continue] dns

and correct the dns entry for your old box.


Regards
Roland

[Fragon]

Hi Matthew,
Would you please change the new box's IP address to a new one(different from these two original IP) using "set_parms ip_address"!
After reboot, you can check if the IP is correct (suppose lan0 for LAN card):
#ifconfig lan0

If session freeze again, that's not the IP duplicate. Or you can boot your old box first, then boot new box!

-ux

[Matthew Pegge_1]

Guys,

We found the set_parms tool inconsistent, which caused more issues, so moved to this method?

The /etc/nsswitch.conf file shows:

hosts: files
services: files
protocols: files
networks: files
rpc: files
netgroup: files

And it makes no difference which order we boot the boxes. Always seems that the last one to boot is the one that we can use. First to boot always freezes!!

[Massimo Bianchi]

Hi,
at what point does the freeze occurs ?


i'm thinking of some things:

- nfs exports/import (check nfsconf)
- custom scripts that do something related to network (adding a second IP on an existing nic..)


Massimo

[Matthew Pegge_1]

The freeze occurs pretty ,uch at the time the 2nd box comes up.

I'm going to attempt this again later when users are off the live server!?! But at moment don't hold out much hope!

[Steven E. Protter]

Few possibilities:
1) A mistake has been made by you.
2) Someone in network administration grabbed that second IP address because they didn't read your project plan.
3) The boxes conflict with each other because one came down and up before the other came down off the network.
4) Syntax error like a dot where it doesn't belong in /etc/rc.config.d/netconf
5) You backed up netconf as netconf.old in the same directory. In 11.00 both files try and load.
I'm leaning toward #3

If the boxes run fine when the other is down then they are fighting with each other.

SEP

[Massimo Bianchi]

Hi,
what is there in the /etc/rc.log and in /var/adm/syslog/syslog.log of the freezing box ?

You can check this now, without a reboot...

Massimo

[Pete Randall]

I like Steven's suggestion #5. Quite often people will back things up with the /etc/rc.config.d directory by copying "blah" to "blah.old". The problem is that everything in the directory gets parsed during startup, so "blah.old" may well undo the changes made to "blah".


Pete

[Enrico P.]

Hi,
are the two server sharing new scsii disks after the modify?
If so, are the termination correctly set?

Enrico.

[RolandH]

Hi Matthew,

can you check, if you have some nfs hard mounts in your /etc/fstab. If yes comment it out. It will often hang if the mount can not be done. So check that you have only local file system mount in your /etc/fstab.

Roland

[Kelli Ward]

Could also be something as simple as a setting in /etc/resolv.conf that is no longer valid with your migration.

GL,
Kel

[Duncan Ball]

Hi Matthew,

Some questions & suggestions you may already have tried...

- Is there more than one interface in either of the systems? Could there be an issue with these?

- Can you get to the console for both boxes? Is the "freeze" only on remote connection?

- Boot up each system separately, and use ifconfig to check (all) the interfaces once all startup activity has completed. This should confirm/clear any possible IP clash.

Duncan

[Matthew Pegge_1]

Thanks to all for your suggestions! We found the problem at last. We had duplicate MAC Addresses, due to someone in the past hard coding this into /etc/rc.config.d/hpetherconf. Changed this to use the default MAC address for the LAN card and BINGO! IT WORKED!!!

Thanks again for all your suggestions!