Hi All.

I am working on bringing back a number of CentOS 7 rigs in our student computer lab back online. No change was made to the existing server machine [running Scientific Linux 6]

Right now there is one remaining thing to resolve: an inconsistency with the rigs’ NIS Clients.

I have configured rcpbind and ypbind following guidance from Server World ( https://www.server-world.info/en/note?os=CentOS_7&p=nis&f=2 ) identically on all of the client machines. I have done this before with previous installs before this. The last time was this summer. Three are behaving as they are supposed to do. Five, however, are not.

In this process I have repeatedly checked that support files ..

/etc/sysconfig/network

/etc/yp.conf

/etc/pam.d/system-auth-ac

/etc/pam.d/system-auth

/etc/login.defs

/etc/sysconfig/authconfig

They are identical.

IP addresses, Netmasks, Gateways, DNS, etc. are correct and all rigs (Server and Clients) do not conflict on any of the machines or /etc/host files.

Disabling firewalls don’t impact the problem.

NFS and other services work fine. All other aspects of networking is fine. You can SSH and access the Server rig from the Client rigs and for locally installed accounts on the clients go the other way. Accounts that leverage NIS cannot log into the Client from the Server or any other remote system.

rpcbind shows that the ypbind/ypserv services are up and Clients and Server, respectfully (and it works on some of the machines).

The rigs that are not working exhibit the following (satisfactory) behavior.

* yptest -u [valid nis username] works with no errors.

* yppasswd will change a password on the NIS server with no errors and other commands like ypchfn will work as well.

* ypwhich, ypcat, ypmap, etc. give the same values we’d see on the NIS server.

…BUT…

* You cannot console-login, ssh, or su into the rigs with valid NIS accounts.

* (The local ypbind -d “debug mode” shows no response to a login, the “secure” log responds to a valid NIS account login with an “invalid user” a the [preauth] phase)

* uids of valid NIS users are not recognized.

* With ls -l, we see the uid of a file’s owner and not the username. (gids are explicitly declared locally on all rigs so they’ll match anyway.)

* The id command for any valid nis user fails with “no such user”

* cd ~[Valid NIS User] responds with “no such directory” even if the home directory exists (and the uid matches the uid on the NIS server).

I am totally at a loss here. I cannot see the difference in what I did on the machines that work and the machines that do not work.

Is there any test that I am missing or are there any files or setting where the culprit(s) on the errant machines may be?

Cheers and Thanks
————————————————————————————-