contents
Next: The Network File System Up: The Network Information System Previous: Using NIS with Shadow

Using the Traditional NIS Code

If you are using the client code that is in the standard libc currently, configuring a NIS client is a little different. On one hand, it uses a ypbind daemon to broadcast for active servers rather than gathering this information from a configuration file. You therefore have to make sure to start ypbind at boot time. It must be invoked after the NIS domain has been set and the RPC portmapper has been started. Invoking ypcat to test the server should then work as shown above.

Recently, there have been numerous bug reports that NIS fails with an error message saying ``clntudp_create: RPC: portmapper failure - RPC: unable to receive''. These are due to an incompatible change in the way ypbind communicates the binding information to the library functions. Obtaining the latest sources for the NIS utilities and recompiling them should cure this problem.gif

Also, the way traditional NIS decides if and how to merge NIS information with that from the local files deviates from that used by NYS. For instance, to use the NIS password maps, you have to include the following line somewhere in your /etc/passwd map:

This marks the place where the password lookup functions ``insert'' the NIS maps. Inserting a similar line (minus the last two colons) into /etc/group does the same for the group.* maps. To use the hosts.* maps distributed by NIS, change the order line in the host.conf file. For instance, if you want to use NIS, DNS, and the /etc/hosts file (in that order), you need to change the line to

The traditional NIS implementation does not support any other maps at the moment.

%


contents
Next: The Network File System Up: The Network Information System Previous: Using NIS with Shadow

Andrew Anderson
Thu Mar 7 23:22:06 EST 1996