YP Access

[MarkSyder]

Hi everybody,

HP-UX server previously running 10.20, now running 11i. I've got it working to the point that I can see everything on the non-root disc and I can log in from my desktop. The users, however, are thrown straight out when they try to access.

I'm pretty certain that yp/NIS is the problem here. They're not actually logging in to this server, but their user accounts are held on it. I've run exportfs to make sure the accounts directory is available and I've been copying files back from a safe area.

Any processes that should be running that I may have missed?

Mark Syder (like the drink but spelt different)

[Robert-Jan Goossens]

Hi Mark,

http://www4.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000062684173

Steps to setup a NIS Master Server, NIS Slaves, and NIS Clients

Regards,
Robert-Jan

[Elmar P. Kolkman]

A simple test to make sure NIS is working is by running the 'id' command for a user that is defined in the NIS files.

[MarkSyder]

Tried both those suggestions, thanks. The id command produces a result. I've followed the instructions in the document and users now get an error message "unable to change to your home directory", so we're a stage further ahead.

Any ideas how to resolve this one?

Mark

[Robert-Jan Goossens]

Mark,

Locale homedir or automount ?

RJ

[MarkSyder]

I'm not sure, but I suspect automount. Some of the home directories are on this server, but some are on others. All are affected.

Mark

[MarkSyder]

New error message that may point someone in the right direction:

I've just tried a yppush from the master server and got the error message "No response from ypxfr on xxxxxxx" where xxxxxxxx is the server I've just upgraded.

Mark

[MarkSyder]

It's working! Thanks for all your help.

It was a sys admin person in our head office who finally came up with the answer, so I'm afraid no one gets 10 points - but your suggestions did get me nearer to the solution, so I'm giving credit where it's due.

Thanks again,

Mark

[Elmar P. Kolkman]

Do you care to share the solution with us? ;-)

[MarkSyder]

I've asked the Sys Admin person concerned to email me with full details of what he did that I hadn't thought of. Once I receive it I will willingly share it with you.

Mark

[MarkSyder]

Here's what I received:

1. The server is a NIS slave but did not have all its maps. Found what maps were missing and used ypxfer to copy them over. The -f says to force the copy:

ypxfr -h (master server name) -f auto.master

only one example attached, but I'm sure you get the picture.

2. Changed the /etc/rc.config.d/nfsconf file thus:

added -C to LOCKD_OPTIONS
added "-l /var/adm/rpc.mountd.log" to MOUNTD_OPTIONS
set NFS_TCP=0
set AUTOFS=0 so the new AutoFS automount will NOT be used (allegedly it is bug ridden).

3. Added lines to the /etc/nsswitch.conf file so it looks like this:

hosts: files [NOTFOUND=continue] dns [NOTFOUND=continue] nis [NOTFOUND=continue] files
services: files [NOTFOUND=continue] nis
passwd: compat
group: compat

it previously had only the hosts line. This lets NFS see the defined port to be used for nfsd as it is not defined in the NIS services map, but only in the file /etc/services.

He then stopped and restarted the NFS mount daemon and all was well with the world (well, the server anyway!).

If this helps someone else do I get my first points :-).

Mark