xinetd : Bad line received from identity server NO-USER

Hi all,

I need some help.
I run an application on a 'mexico' computer, which rsh to many others
computers to run jobs, using rsh and rcp commands. All works fine, but
sometime, after hours of jobs and thousands of successful rsh and rcp
commands, the 'mexico' computer can't rsh to the 'rio' computer.
There is no particular event logged on the 'mexico' computer.
On the 'rio' computer, log message until the faulty rsh is :

Sep 13 20:07:13 rio pam_rhosts_auth[8462]: allowed to gregoire@mexico as
gregoire
Sep 13 20:07:13 rio rsh(pam_unix)[8462]: session opened for user
gregoire by (uid=0)
Sep 13 20:07:13 rio rsh(pam_unix)[8462]: session closed for user gregoire
Sep 13 20:07:18 rio pam_rhosts_auth[8488]: allowed to gregoire@mexico as
gregoire
Sep 13 20:07:18 rio rsh(pam_unix)[8488]: session opened for user
gregoire by (uid=0)
Sep 13 20:07:18 rio rsh(pam_unix)[8488]: session closed for user gregoire
Sep 13 20:07:18 rio pam_rhosts_auth[8515]: allowed to gregoire@mexico as
gregoire
Sep 13 20:07:18 rio rsh(pam_unix)[8515]: session opened for user
gregoire by (uid=0)
Sep 13 20:07:19 rio pam_rhosts_auth[8540]: allowed to gregoire@mexico as
gregoire
Sep 13 20:07:19 rio rsh(pam_unix)[8540]: session opened for user
gregoire by (uid=0)
Sep 13 20:07:26 rio rsh(pam_unix)[8540]: session closed for user gregoire
Sep 13 20:07:41 rio xinetd[8572]: Bad line received from identity server
at 192.168.1.80: 563 , 514 : ERROR :NO-USER

(mexico computer is 192.168.1.80)

'mexico' computer is a Linux Kernel 2.6.9-34 RedHat Enterprise 4, two
processor Opteron 246 and 16gb ram. My running application is 32bits.
'rio' computer is a Linux Knerel 2.4.21-4 RedHat Enterprise3, one
processor Duron and 2gb ram.
And the Nis server is a Sun running Solaris 2.8. Nis server was recently
patched with the latest update because large amount of request crash the
ypserv service. Now it seems to work correctly.

Does anyone can give me a clue to find what's wring ?

Thanks for your help.

Grégoire Avot.

On Fri, 14 Sep 2007, in the Usenet newsgroup comp.os.linux.networking, in
article <46ea4bc8$0$14829$(E-Mail Removed)>, Gregoire Avot wrote:

>I run an application on a 'mexico' computer, which rsh to many others
>computers to run jobs, using rsh and rcp commands. All works fine, but
>sometime, after hours of jobs and thousands of successful rsh and rcp
>commands, the 'mexico' computer can't rsh to the 'rio' computer.

Wow - haven't seen rsh and rcp used in a long time. The fact that you
are also running NIS means you are aware there is virtually no security
in these protocols.

>There is no particular event logged on the 'mexico' computer.

Not even an error message that the 'rsh' failed?

>On the 'rio' computer, log message until the faulty rsh is :

>Sep 13 20:07:41 rio xinetd[8572]: Bad line received from identity server
>at 192.168.1.80: 563 , 514 : ERROR :NO-USER
>(mexico computer is 192.168.1.80)

You'll have to look there - the identity server says that the user does
not exist.

>'mexico' computer is a Linux Kernel 2.6.9-34 RedHat Enterprise 4, two
>processor Opteron 246 and 16gb ram. My running application is 32bits.

How busy is it?

>'rio' computer is a Linux Knerel 2.4.21-4 RedHat Enterprise3, one
>processor Duron and 2gb ram.

while very old, it's probably not the cause of the problem.

>And the Nis server is a Sun running Solaris 2.8. Nis server was recently
>patched with the latest update because large amount of request crash the
>ypserv service. Now it seems to work correctly.

None the less, that could be the problem. I'd run a packet sniffer to
see if NIS is the problem - the 'mexico' system being unable to get the
right identity tokens.

Old guy

Thanks a lot for your response.
I found the problem : since each rsh/rcp request open a new socket which
stay in "TIME_WAIT" for two minutes, the computer ran out of free port.
This causes random problem (nis failure...)
I solve the problem by writing a tcl script which handle file transfer
and run command with a persistant connexion between computers. Now
everything works well, and faster !