system hangs at login prompt

[Adam Noble]

I realise this is quite a vague subject matter, however would be grateful for any ideas. In the last two weeks two of our servers have exhibited the same problem, whereby when using telnet, remsh etc you can enter a username and then it simply hangs and eventually times out. All users on the system experience no degredation in performance.

I have never seen this before, however I have seen the system hang numerous times before you reach the login prompt. I can ftp to the system and have altered cron to run a number of perf stats on the system which come back without any obvious issues. The console is showing errors relating to bad network connection to the Hub, however I'm unsure if this is connected as we can obvioulsy ping the server. The console however on both occasions has completely locked up. Any ideas or suggestions would be grateful.

[RAC_1]

are there any errors being logged in syslog file.

if loggin is not enableddo it with inetd -l so that it logs any error messge in syslog.

[Stefan Farrelly · Accepted solution]

Its almost certainly a network/dns resolve issue. At some point dns lookups fail due to a dns server being unreachable or not responding or faulty network connection/card.

Things to try are;
1. configure /etc/nsswitch.conf to look at /etc/hosts (files) also if it cant resolve to your dns server.
2. add in roots home dir a .nslookuprc file with retry=1 timeout=3 in it (or something similar) so that dns lookups timeout faster if your dns server is unreachable or doesnt respond
3. check you network card and cable connections to ensure theyre all good. Check network errors and collisions (use lanadmin or netstat -s as if youre getting lots of either this may prove you have a physical network problem).

[T G Manikandan]

You have to first check the /var/adm/syslog/syslog.log file for any messages.

Make sure that your switch and the NIC on the server are running at same duplex settings and are of similar speed.

you receive these messages of bad connection as the speed and duplex settings are different from NIC on the server and the port on the switch side

[Jean-Louis Phelix]

hi,

You can find a lot of related things in previous posts, but this problem is very often due to :

- NFS home mounted homes. The work-around is to change your history file (HISTFILE in your .profile) to something local (ie /tmp/.$LOGNAME.histfile

- problems with quota checks at login

Regards

[RAC_1]

was just going to tell to check nfs mounted home directories as JEan told.

Can you give more details about syslog errors and network settings on switch and on NIC?

[harry d brown jr]

post the outpiut of lanadmin - select lan - select display

Now, what changed TWO WEEKS ago?? Did someone change any settings on your network router/switch? Did you experience a power outage? Install new software? What OS and release are you running?

live free or die
harry

[Adam Noble]

Yes I checked syslog and there were no obvious errors!After the server booted everything is fine again! Is there some sort of reverse DNS lookup when you login that occurs. I should say however that currently the /etc/nsswitch.conf file is set to use the servers hostfile.