Original Question:
Systems running SunOS v4.1.3_U1 are configured to run under one NIS domain.
Yesterday, two of the NIS clients were generating the error message:
NIS: server not responding for domain "noname"; still trying.
After cleaning the /var/yp files on all the clients and rebooting the
systems, they are no longer generating this message. Today, the the two NIS
Slave Servers are the ones generating the error message described above.
History:
I have two subnets under the same NIS domain. Each subnet had its own
NIS Master and Slave servers (I am working on keeping only one Master). I
moved seven SPARC stations from one subnet to another. One of the
systems I moved used to be the NIS Master Server at its old location and
two of them were NIS Slave Servers. I am trying to setup these as the
Master and Slave servers at their new location and the original NIS
Master and Slave servers as just NIS clients.
On my first try, two of the new clients were generating the error message
described above. After removing previous references to NIS (cleaned
/var/yp) and restarting ypbind on the clients, I was able to clean the
clients yesterday. Today, I came to the office and the two NIS Slave
servers are the ones generating the error message.
I am at a lost. We are also running into many rpc timeout errors. Any
suggestions?
Thanks to:
Lee Ann Goldstein <lgoldste@liquid.ladc.lockheed.com>
Ernie Chan <chane@eis-e1.comm.mot.com>
Glenn Satchell - Uniq Professional Services <Glenn.Satchell@uniq.com.au>
Suggestions:
Oooh, ow! There's your first problem- you should only ever have one NIS master
per domain. (and you really only need one slave per subnet)
--- I went ahead and kept only one Master but the problem continued.
I recommend that you get a copy of O'Reilly's "Managing NIS and NFS", which
gives clear instructions on how to do what you want to do.
It sounds like you just haven't configured /etc/defaultdomain correctly to
contain the text of your domainname. If these machines have been rebooted
recently and this file contains "noname" then that becomes your domainname and
it tries to look for NIS services from a server of that domain...
--- All of them had the correct NIS domainname.
The quickest fix is probably just to re-initialize master and all the
slaves, ie /usr/etc/yp/ypinit. The slave servers remember who the
master is and will only accept updates from that machine.
--- This almost worked.
Solution:
I re-initialized all the NIS servers but the error came back. Eventually,
I killed ypbind on all clients and ypserv on all servers. Then, processed to
re-initialize all the NIS servers and restart all the clients. That cleaned
NIS. Everything is back to normal.
The rpc timeout errors went away as soon as the network was normalized.
/---------------------------OOo--oOO-----------------------------\
| Ricardo J Ruiz E-MAIL: rruiz@census.gov |
| U.S. Census Bureau |
| Technical Services Division FAX: (301) 457-2477 |
| Room 1435-3 |
| Washington, DC 20233-2300 |
| |
| Visit our award-winning home page: http://www.census.gov/ |
\---------------------------ooo-ooo------------------------------/
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:25 CDT