CDE/Network problem

[Basil Holloway]

Have a 7011-250 with AIX 4.3.3.0
Install AIX and can log onto CDE desktop.
Install onto network via smit/communications/TCP-IP/Min/en0 and cannot log
on to CDE desktop. 7011-250 can ping router and an Win XP machine. Win XP
machine can ping 7011-250.
Run " #ifconfig en0 detach " and can log onto CDE desktop.
Tried installing network through configassist TCP/IP but then cannot log
onto CDE. Take it out with " #ifconfig en0 detach " and can log onto CDE
deaktop.
Would appreciate any pointers to solve apparent Network set up and CDE logon
conflict.
Thanks

[sbzx]

Probably a name resolving issue. Try the following:

(1) Make sure /etc/hosts has the correct IP address and hostname for
your host.
(2) Make sure both IP address and hostname resolve correctly ("host
<IPaddress>" and "host <hostname>" should give you the same output)
(2a) You may want to check /etc/netsvc.conf, you should probably see a
line similar to "hosts=local,bind" for DNS, or "hosts=local" or
nothing for no DNS.
(3) Reboot your system (it may not be necessary but it's better to
start with a clean record).

I hope this is helpful. If not, post exact error message and TCPIP
setup details.

[Basil Holloway]

sbzx thanks for your reply and clear instructions

> (1) Make sure /etc/hosts has the correct IP address and hostname for
> your host.

running #smit/hosts/list all hosts gives
127.0.0.1 loopback localhost # loopback lo0 name/address 192.168.0.166

> (2) Make sure both IP address and hostname resolve correctly ("host
> <IPaddress>" and "host <hostname>" should give you the same output)

running #host 192.168.0.1 gives local host is 192.168.0.166
running #host localhost gives local host is 127.0.0.1

sbzx, as you anticipated, there is an error here.
Could you please tell me how to correct it.
Trying to learn. Where did 127.0.0.1 come from. Went through my set up notes
and I did not input this address.

> (2a) You may want to check /etc/netsvc.conf, you should probably see a
> line similar to "hosts=local,bind" for DNS, or "hosts=local" or
> nothing for no DNS.

running #find / -name netsvc.conf -print the computer found nothing.

> (3) Reboot your system (it may not be necessary but it's better to
> start with a clean record).

Just an old retiree learning AIX. Old Windows habit, I always reboot on
major changes. The old 7011-250 takes about 6 to 7 mins to reboot, but it
gives me time to write up my notes. Is a root logout and login the same as a
#shutdown -r ?

> I hope this is helpful. If not, post exact error message and TCPIP
> setup details.

running #smit /communications/tcpip/min config/en0 settings show

Host Name localhost
Internet Address 192.168.0.166 (checked my Workgroup
and no computer, router, switch, hub
or
printer was using this address so I made it up)
Network Mask 255.255.255.0
Network Interface en0
Nameserver Internet address is blank ( On
setup I input 192.168.0.1, my router but AIX knocks it out)
Nameserver Domain Name is blank ( On setup
I input basil.com but AIX knocks it out)
Default Gateway 192.168.0.1 ( My
router address )
Your Cable Type is blank
( On setup I input dix but AIX knocks it out)
Start Now no

sbzx, there is never any error message. With network active, at root log
on, the Blue CDE screen comes up, stays for 4 to 5 mins then goes black. I
have then waited up to 20 mins before pushing the reset button. On both the
blue and black screens, the mouse curser is responsive to mouse movements,
but is allways an hourglass.

When you give me the solution to (2) above, as the computer has had run
#ifconfig en0 detach
do I run
#ifconfig eno up
before or after I do you corrections ?

Thanks
Basil

[sbzx]

Basil,

There seems to be a number of problems with this setup. Here's a way
you may want to try to correct them:

(1) It is unhealthy to have two "localhost" lines in /etc/hosts. Find a
nice hostname other than "localhost" and set it as your hostname using
"smitty hostname". When done, clean up /etc/hosts, it should look
something similar (you set "sweetie" for hostname):

localhost 127.0.0.1
sweetie 192.168.0.166

(2) Since you have no working DNS server that resolves your FQDN,
remove (or rename) the /etc/resolv.conf file. This practically
eliminates the need to have a valid /etc/netsvc.conf, so we can forget
about that.

(3) Reboot. Normally this would not be necessary, AIX is not Windows, a
restart of X11 would be enough, but I am no X11 expert I don't know
which processes should be killed and restarted. I never use X11. :)

And no, logout and login root is very far from being the same as a
"shutdown -r". The former just reloads the user settings of root, while
the latter reinitializes the entire OS including devices.

[Uli Link]

Hi Basil,

When logging into CDE the machine tries to resolve/verify it's IP
address to it's hostname. Functional name resolution is mandatory for
CDE. You can start a Failsafe X11 desktop session for running X11 smit.

AIX defaults to DNS resolution, so if your DNS does not resolve the
hostname of your 250 workstation CDE will hang as you described.

You can overwrite this behaviour by advising AIX to check a local
/etc/hosts file before querying DNS.

If /etc/netsvc.conf does not exist (the default), create it with a
content of
hosts = local, bind

127.0.0.1 is a loopback IP address. Packets with this destination will
never leave the machine. Nothing wrong. Every UNIX box and every router
has this loopback IP address internally.

Don't need to reboot. You can reset the X11 server from the login screen
with --> "Options" --> "Reset Login Screen".

--
Uli

[Basil Holloway]

Thanks sbzx and uli, but on trying your suggestions every time the desktop
came up it was overlaid with a window titled Action Required saying " The DT
messaging system could not be started. To correct the problem
.................. correct /etc/hosts ...................see DT users
Guide"
I think this users guide is GC23-2432 but cannot find a copy on web.
Anyway this Action Required window has gotten me in an endless loop between
an OK button and the logon box.
Wiping system now and will do a fresh install using a new host name leaving
2 Nameserver fields blank & cable type as N/A.
Will let you know how I get on tomorrow.

[Claudiu Costin]

Post to the list your:
/etc/hosts
/etc/netsvc.conf
/etc/resolv.conf
and the output of the following commands:
egrep 'dtspc|cmsd|ttdbserver' /etc/inetd.conf
host $(hostname)

[Basil Holloway]

Claudiu here is that info you wanted

> Post to the list your:

> /etc/hosts

127.0.0.1 oldage

> /etc/netsvc.conf

don't have this file

> /etc/resolv.conf

don't have this file

> and the output of the following commands:

> egrep 'dtspc|cmsd|ttdbserver' /etc/inetd.conf

line 1
ttdbserver sunrpc_tcp tcp wait root /usr/dt/bin/rpc.ttdbserver
rpc.ttdbserver 100083 1
line 2
dtspc stream tcp nowait root /usr/dt/bin/dtspcd /usr/dt/bin/dtspcd
line 3
cmsd sunrpc_udp udp wait root /usr/dt/bin/rpc.cmsd cmsd 100068 2-5

> host $(hostname)

oldage is 127.0.0.1

Thanks
Basil

[Basil Holloway]

> Will let you know how I get on tomorrow.

Could not wait till tomorrow.
Reinstalled AIX 4.3.3.0 and in configassist screen accepted paging space
increase and set root password.
Logged on as root . Screen said "Welcome localhost"
Added hdisk0 and hdisk2 to rootvg along with hdisk1
Increased the PP's to hd5,hd8,hd4,hd2,hd9var,hd3 and hd1
Altered the FS's in /,/usr,/var,/tmp and /home
Altered the host name from localhost to oldage
Rebooted and got the Action Required window again re /etc/hosts and the
endless loop
Went to command logon and in
#smit/hosts/list all hosts got 127.0.0.1 loopback localhost # loopback (lo0)
name/address
#smit/hosts/remove a host and knocked out 127.0.0.1
#smit/hosts/add a host and entered 127.0.0.1 as internet address and oldage
as host name
#smit/hosts/list all hosts got 127.0.0.1 oldage
rebooted logged on as root Screen said "Welcome oldage " and the Action Reqd
screen did not appear.
As it is after 10.00 pm, will tomorrow
(1) set up printer
(2) install ver 11 update
(3) set up network
(4) install Network Communicator 4.8

Thanks guys

[Tim Clarke]

"Basil Holloway" <FAKE...@bigpond.com> wrote in message
news:%4GFf.244043$V7.1...@news-server.bigpond.net.au...

>
> Rebooted and got the Action Required window again re /etc/hosts and the
> endless loop
> Went to command logon and in
> #smit/hosts/list all hosts got 127.0.0.1 loopback localhost # loopback
(lo0)
> name/address
> #smit/hosts/remove a host and knocked out 127.0.0.1

Hi Basil,
the above step is *not* a good idea, all *NIX and other system's TCP/IP
stacks have a (virtual) loopback adapter (lo0) and (alias) hostname of
"localhost" set up on that interface. It is a "well-known name" which *must*
exist on that IP address, which is associated with the loopback adapter, and
it is often used for intra-system communication.

> #smit/hosts/add a host and entered 127.0.0.1 as internet address and
oldage
> as host name

If you want "oldage" as the hostname for "this" system, then you need to
configure its correct IP address (e.g. a local one of the form 192.168.n.n
with a subnet mask of 255.255.255.0 - IP Class C address subnet mask) on a
physical/real adapter (en0/et0) and associate the "oldage" hostname with
that IP address, either manually. via /etc/hosts, or a local DNS server.
This is basic TCP/IP setup/configuration.

HTTH
--
Regards,
Tim Clarke (a.k.a. WBST)
Guildford, U.K.

[Jose Pina Coelho]

Uli Link <VonRechts...@usenet.arcornews.de> wrote in news:43e5c550$0
$348$9b4e...@newsread2.arcor-online.net:

> If /etc/netsvc.conf does not exist (the default), create it with a
> content of
> hosts = local, bind

Unless your local DNS is capable of *correctly* resolving IPv6 addresses,
use hosts=local,bind4 instead, otherwise it will try to get IPv6 addresses
before IPv4 and will hang on "netstat -r" and "netstat -i" trying to
resolve "::1" (IPv6 loopback).

--
Doing AIX support was the most monty-pythonesque
activity available at the time.
Eagerly awaiting my thin chocolat mint.