Communication Problems between Database Servers

[Adrian Sobers2]

I'm having problems communicating between two of our database servers.

Actually I can ping and tnsping from TEST to PRODUCTION...but I cannot ping or tnsping from PRODUCTION to TEST.

Attached is a zip file containing the output from netstat -rn from old and new servers.

old.txt --> output fron netstat -rn on TEST

new.txt --> output fron netstat -rn on PRODUCTION

Any help would be greatly appreciated. I suspect it is something to do with my routing information/tables. But networking is not up my steet period. Thanks... ;)

[Adrian Sobers2]

oops, zip file ;)

[Sanjay_6]

Hi,

No Firewalls in between ?.

Can you ping test from production using ip_address (just OS ping).

Hope this helps.

Regds

[Geoff Wild]

Well, for one thing, you have no default gateway on test....

Other things to check - DNS resolver - do you do files then dns? check nsswitch.conf.

If files first, are both servers in each other's hosts files and have the correct ip?

Also check the netmask for your primary interfaces - are they the same?

Rgds...Geoff

[Adrian Sobers2]

Attached is the nsswitch.conf from production. Yes the hosts files are correct on both servers containing correct IP addresses and hostnames.

[Ron Kinner · Accepted solution]

I think what you have is an addressing conflict. If you look at your TEST netstat you will see that it has two addresses:

200.200.120.3 200.200.120.3 UH 0 235 lan0 4136
200.200.120.8 200.200.120.8 UH 0 188154 lan1 4136

(This, by the way, is an unsupported configuration. Having two interfaces on the same LAN is sort of frowned upon by HPUX and may or may not work.)

If you look at PRODUCTION you see only one address:

200.200.120.3 200.200.120.3 UH 0 lan1 4136

but it's the EXACT SAME ADDRESS as TEST's lan0. So I expect that what is really happening is that a ping from Test to Production causes a ping with source 200.200.120.8 to be sent to 200.200.120.3 but since 200.200.120.3 is a local address it may not even reach the wire but even if it does it will be sent to the MAC found in the arp table (arp -a) for 200.200.120.3 which will most assuredly be that of TEST's lan0 and not that of PRODUCTION. TEST's lan0 will happily answer the ping by replying with a packet to 200.200.120.8. So it works.


If you ping from PRODUCTION to TEST the ping will reach the TEST, I assume on 200.200.120.8, but the Echo Reply will be sent to its own MAC for 200.200.120.3 so PRODUCTION will not get a reply to its ping. So it fails.

You could use Ethereal to prove it but I'm pretty sure that is the problem. Remove the 200.200.120.3 from TEST and things should start working.

Ron

[Adrian Sobers2]

Ron,

Thanks, but how do I remove 200.200.120.3 from TEST?

[Robert Steininger_2]

ifconfig unplumb

for example: ifconfig lan1 unplumb

[Robert Steininger_2]

in your case:
ifconifg lan0 unplumb

[Adrian Sobers2]

I did this, but still having problems. I cannot ping 200.200.120.8 (TEST) from production (200.200.120.3) still.

Also my HummingBird Exceed client to connect to TEST is messed up...

going into meeting...

[Robert Steininger_2]

stupid question: did you check your calbe connection? is the lan1 port connected to the hub/switch??

the command "landiag -g 1" should display something like that:

......
Administration Status (value) = up(1)
Operation Status (value) = up(1)
......

[Adrian Sobers2]

I get the output "up" for both of the things/categories in your last post for the command.

I checked the network and it is plugged in to a connection point on the wall. That connection point is dropping off wall physically but it is plugged into the SOCKET itself...what do you think?

I'll check it in the hub/switch now. I have like ZERO networking experience :(

Also I changed the IP address from 200.200.120.3 on test to 200.200.120.8...the test machine had the same IP as PRODUCTION in the hosts file. Do I have to do a reboot or a net/start stop.javascript:postAnswerSubmit('submit');
submit

[Adrian Sobers2]

Also, when I try to do a net stop/start, here is the output:

root:INLAND3>/sbin/init.d/net stop
root:INLAND3>/sbin/init.d/net start
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
/sbin/init.d/net[118]: ^M: not found.
interface: ifconfig: no such interface
interface: ifconfig: no such interface
ERROR: Failed to add route entry because its interface is not
yet initialized. May need to add this route entry with
a route commad after the interface is up :
: unknown command


getting deeper into shark infested waters it seems....arghhh

[Victor BERRIDGE]

Hi Adrian,
Before you try to reboot anything, go and doublecheck AND correct the values in /etc/rc.config.d/netconf file - that is remove the unwanted ip and lan config

then you can reboot and see

All the best
Victor

[Adrian Sobers2]

Attached is a zip file containing my netconf file on 200.200.120.8 (INLAND3) and the output from the command lanscan

Does it seem correct? I cannot still ping TEST maching from production...it does not recognize it...Do I have to reboot?

[Victor BERRIDGE]

Who is inland3 test or prod because you have in your netconf the indentical conf for lan0 and lan1 but one is down

Difficult to say not knowing who is who

[Adrian Sobers2]

Test is 200.200.120.8 "INLAND3"

Production is 200.200.120.3 "inland8"


Do I have to change anything according to what I have told you?

[Robert Steininger_2]

check your /etc/rc.config.d/netconf file the! may you have unwanted control characters in there ("^M")

the command "cat -v /etc/rc.config.d/netconf"
should not display any "^M" characters!

[Adrian Sobers2]

Yes I think those characters were from my saving it in wordpad and ftping it back over :( oops..is there a way to remove them all in one fell swoop?

[Robert Steininger_2]

dos2ux /etc/rc.config.d/netconf > /etc/rc.config.d/netconf.new

[Victor BERRIDGE]

ux2dos netconf>titi
cp titi netconf
And check the perms before/after

All the best
Victor

[Victor BERRIDGE]

Sorry I was on the phone... dos2ux

[Victor BERRIDGE]

OK so if this netconf is the one of the test box if lan0 is wired why not keep this as default...

What do you have on the procution box?

[Adrian Sobers2]

Victor,

Attached is the netconf from the production machine: inland8 (200.200.120.3)

[Victor BERRIDGE]

You ar missing in your netconf
ROUTE_GATEWAY[0]=""
This is if you wish to use lan0...
This should be configured like if you were on the lans here 200.200.120.1


Waiting to see the other netconf...

All the best
Victor