7028 6C4 problem on login 2

[antonioelderprado]

Hi sys admins,
I have a 7028 6C4 machine (actually 3 in a rack) that few times when i try to telnet to the machine from another unix box the login stops before connected to aixtest08,
Escape character is '^]'
and hangs there...

when I got to the console, the machine cannot communicate to any other box; however if I ping from another machine it comes back without any problem.

I needed to reboot the box.

Do you have any idea or clues of what is going on.


This happens on two machines of same type 7028 6c4 that are racked together. Could this be a temperature issue, it is very hot near the boxes. ?


thank you all for your help.


./antonio/.

 

[aixnusr]

the problem may be either because of NIS/DNS ( if its user ) or may be of network card config ( check auto or full duplex) or of over loaded.. its not clear from your text when u r experiencing the problem... we have >50 7028-6C4 systems and 8 are in same rack and still no such pblm

 

[antonioelderprado]

one more piece of info:

when I do a netstat -D
on the part of IP TCP and UDP gets me:

Ipkts Opkts Idrops Odrops

IP 54083848 45708985 8019895 8019794
TCP 6523840 8447001 13 0
UDP 39164649 34522104 21 0


do you thinks is a routing problem (IP).,

 

[antonioelderprado]

some more info...

actually I haven't fixed yet.
The problem is I needed to get the machine back on the network fast and I rebooted. When rebooting all goes away.

I believe it is not dns issue because it did try to telnet using i.p. instead of name and the result was the same.

the problem is really strange.

I can ping from other machine to the problematic box.
I cannot telnet to the problematic box.

from the console, I cannot ping or telnet to any other machine.
I cannot ping even to the machine itself.

funny, uh!.

 

[aixmurderer]

Did you check your paging space on the problem box? Sounds like a lack of PS issue. (lsps -s)

IBM Certified Confused - MQSeries
IBM Certified Flabbergasted - AIX 5 pSeries System Administration
MS Certified Windblows Rebooter

 

[antonioelderprado]

hi Aixmurderer ,

I am inclined to think so, however now: the paging is# lsps -s
Total Paging Space Percent Used
1792MB 1%
#

pretty low.

is it possible that whenever something internal (paging, kernel, tcp stack) fills up automatically kills inetd or some services ?


thank you murderer... (cool nick).

./antonio/.

 

[needcoffee]

Can you ping localhost from the problem machine? What is the output of netstat -C from problem box? Is the telnet daemon running on box? Is inetd active (check with lssrc)?

What type of ports are the network cards connected to, dump hub, managed hub, managed switches? How are they configured?

Is this problem just occuring or has it happend before?

needcoffee

 

[antonioelderprado]

needcoffee...

telnetd is working
inetd is active
netstat -C respond with the correct routing tables.

the machine is connected to dumb hubs, those drops that comes under the desk (two d-lines), I did connect one of them to a fry's hub - 4 port and connected the 3 aix boxes to that hub. Do you believe it is too much for that single landrop ?

thank you needcoffee, ... me too.

./antonio/.

 

[farhsyed]

The same has happened on F50 machines with me. You need to check the ethernet adapter speed and change it to 100 Full duplex instead of autonegotiate and then boot the machine.

sure this will resolve the problem.

Farhsyed

 

[antonioelderprado]

thank you all who kindly send me information.

I did change the lan from autoneg to fullduplex and so far no problem.

once again,

thank you all,


./antonio/.