Hi !
The orig. posting was:
-----------------------
We are running a system with Solaris 2.5, at all client and server machines.
Last night we reconfigured a direct map for the automounter -- the direct map
as well as the other maps are distributed through NIS. Our goal is to export
a central /var/news directory to all NFS clients.
We re-made the /var/yp stuff on our NIS server. We NFS-shared the /var/news
directory on the server machine.
To avoid a potential deadlock (cycle) in case the /var/news server
reads the maps through NIS, we set up a direct map for a mountpoint
/var/newsexport and let the client /var/news directory be a link
to /var/newsexport.
The next step was to restart the automounter at the clients.
We used the scripts in /etc/rc2.d for autofs, which would kill the automountd
process and then restart it, followed by the 'automount' command.
However we encountered some problems with the NFS-clients:
Of course, most of the NFS dirs were still mounted i.e. "busy".
Restarting did not succeed printing the following diagnostics:
automountd[27318]: svc_tli_create: could not bind to requested address: No Error
automountd[27318]: svc_create: svc_tli_create failed
automountd[27318]: svc_tli_create: could not bind to requested address: No Error
automountd[27318]: svc_create: svc_tli_create failed
automountd[27318]: svc_tli_create: could not bind to requested address: No Error
automountd[27318]: svc_create: svc_tli_create failed
automountd[27318]: unable to create service
There is a problem with rpcbind here it seems. Even worse, the automounter did
not come up correctly, leaving the client in a braindamaged state !
Only rebooting would fix this :(
My question:
How can we avoid rebooting -- or better, how should we reconfigure and
restart the automounter ?
Besides, are there any inferences with automounting and cachefs ?
-----------------------------
The answer(s):
For Solaris 2.5 it is sufficient to issue the 'automount' command.
As an alternative one could use indirect maps instead of direct maps which
would avoid to restart automount.
A nice trick to avoid mount loops -- besides proxy sym-links --
is to use the -null map option.
Actually, our real problem was that we accidently referred to oudated manuals
on Solaris 2.0/2.1 :(
Thanks to all that replied:
Frode Stromsvag
Casper Dik
Marc S. Gibian
Anthony Worrall
Kevin Davidson
and others (sorry if I missed someone).
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:18 CDT