SUMMARY: Clients hang after server reboot

From: Michael Stumpf (
Date: Wed Dec 08 1993 - 09:03:26 CST

I received three answers so far, and the problem seems to be well-known
and solvable. The main point is that the clients hang because they
loose connection to the NIS server (if the NFS server also is the one
and only NIS server, which is the case here), and cannot reestablish
the connection because the "old" port on the server is not in use any
So, there is one easy fix: Run a second (slave) NIS server.
Another fix (Richard Czech, is, to force the client to
bind to the new port on the server. The client binds that way
automatically, if the command issued involves a sufficient time-out
value. rlogin does not and terminates, before the client tries to
bind. Surprisingly, rsh xterm does, as Richard pointed out. He
suggests the following:

        server% xhost client
        server% rsh client xterm -display server:0

(Needless to say that the server needs to run an X-server and that this
procedure must be repeated for every client haning...)

James also mentioned patch 100342.

Thanks to all who responded:

Here comes the original posting:
>Hi folks,
>once in a while, as usual, my server has to be rebooted for some
>reason. Just a shutdown or fasthalt and reboot. Sometimes, the diskless
>clients do not recover from this but hang. The user sees the login
>prompt, types in his userid but is never prompted for his password. The
>client does honor ping request, but does not answer rshs nor rlogins -
>it has to be rebooted, too. This is not always the case, but too often.
>Can anyone shed light on this?
>Server is SS10 Model 412 running 4.1.3; clients most affected are
>ELCs (IPXen often are not while ELCs are...).


| Michael Stumpf, Psychological Institute, University of Freiburg              |
|                 Niemensstr. 10, D-79085 Freiburg im Breisgau, Germany        |
| RFC822:       NIC-Handle:    MS59 |

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:31 CDT