Getting logged out after correct login and passwd

[miracle]

I have an old D370 and after the last two reboots I cannot log in for many hours after.
HPUX 11.00 This system has not been patched for some time but was working okay before.

inetd is okay, I can still ftp and I can use rexec. There is no messages in syslog that pinpoint a problem.

When I attempt to login
telnet to system
Login:
Password:
Both are okay, then I get logged out. Any clues?

[Bharat Katkar]

HI,
I doubt whether you had anything wrong in /.profile or /etc/profile. Try booting the server into single user mode and check both this file.
For booting into single user mode:
1. Halt the Server at PDC prompt
2. Main menu> bo
3. Do you want to interact with IPL> yes
4. ISL> hpux -is
# more /.profile
# more /etc/profile

Hope that helps.
Regards,

[Bharat Katkar]

Hi,
Also check /etc/passwd file for valid shells for all users.

Regards,

[miracle]

I am not in a position to boot the system. First it is production, but users use a client to attach to the application on the system. There has been no changes to /etc/profile or are there any strange shells. Last week we were able to login after a time. As mentioned I can run rexec commands as root with no problems.
Any other suggestions?

[Bharat Katkar]

HI,
Okay then.. do you have a console and did you get a login prompt there. Try logging as root from console. See if it work... if so then try doing telnet to itself.

# telnet hostname

In case you don't get login prompt here then hopefully your telnetd daemon is not working...


Regards,

[Bharat Katkar]

sorry ignore that i just forgot that you r getting login... anyway i am just stucked now.

[miracle]

I am not on the site so a console is not going to help. As mentioned, I can telnet from any host. I get the login prompt, I put in my id and when the password prompt appears, I enter my password, (which is correct) then I get a message immediately saying "Connection closed by foreign host" and the session has dropped out back to the system I was issuing telnet from.

[Bharat Katkar]

HI Miracle,

Since this is a remote machine, could be possible that no. of telnet sessions have exceeded its limits.

Try checking the following Kernel Parameter:

npty : Number of pseudo TTY device files
nstrtel : Number of telnet session device files


Just a thought.
Regards,

[miracle]

There are no users currently logged in.

# rexec serverx w
I put password and there is no one logged in.

When I have a look at /usr/sbin/dmesg the only thing of note is
"Equivalently mapped reserve pool exhausted;
Overall application performance may be improved by increasing the "eqmemsize"

# rexec serverx /usr/sbin/dmesg

Would this cause this problem?

[Bharat Katkar]

What is the value you have used for eqmemsize

[miracle]

it's set to default of 15

[Michael Tully]

What other changes have been made to the system recently?

Additional software ?
More software load ?
Any changes in hardware?

[Michael Tully]

Seeing you can do rexec's, can you run an xsession from the host from a PC, like reflection X or exceed or an equivalent?

[Oved]

are u using some kind of security tool ?

Maybe it denies you from doing a login by accident.

Is there a firewall that may cause you to disconnect ?

I hope it helps,
Oved

[florence mathon lermusi]

hi,

does the problem appear with different login account ?

if you are using ksh, try to purge your .sh_history file (just make a touch on it), i already see such a case with a corrupted one...

[Michael Tully]

As I'm a caring and sharing person I've taken over this thread from my collegue. Up to this morning I didn't know this was his ...

The facts so far:

Couldn't login with any account using telnet
Could use ftp and rexec
Was able to start and use a CDE session booting it from this server.

After further examination it was revealed that there were a few emails stuck in the system. I run an 'elm' email and all of a sudden telnet was again working. There were also some disk related messages in syslog, and the diagnostic daemon was complaining about excessive disk I/O. We ended up placing a call with the local HP response centre. The suggestion is we have some sort of problem with an older version of telnetd. We will be patching this particular system on the weekend. We will update this call after then, thank to everybody who has participated thus far. Points to be awarded. Hopefully my collegue will do the right thing and award me none ... for not helping sooner.

[miracle]

That's right Sir Tully should get nothing besides I think he has enough points anyway with all that green growth ... but I didn't tell him there was a problem either.

The latest:
We have updated to March 2003 patch bundle
Problem still persists, but with HP support have looked at 'eqmemsize' as being the possible culprit. We have a change in the system for this weekend,to hopefully fix this once and for all.