Re: NIS and autofs and /net
From: P.T. Breuer (ptb_at_oboe.it.uc3m.es)
Date: 02/21/04
- Next message: Michael Heiming: "Re: LTSP"
- Previous message: Jean-David Beyer: "Re: large binary immediately SEGV's"
- In reply to: Rick Denoire: "Re: NIS and autofs and /net"
- Next in thread: Rick Denoire: "Re: NIS and autofs and /net"
- Reply: Rick Denoire: "Re: NIS and autofs and /net"
- Reply: Rick Denoire: "Re: NIS and autofs and /net"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Sat, 21 Feb 2004 20:03:35 +0100
Rick Denoire <100.17706@germanynet.de> wrote:
> ptb@oboe.it.uc3m.es (P.T. Breuer) wrote:
>
>>Rick Denoire <100.17706@germanynet.de> wrote:
>>> I incorporated a Linux client into a Sun NIS domain, "seems" to work.
>>> According to the Redhat instructions (using RH AS 2.1), one has to add
>>> a line to the client's /etc/passwd file like this:
>>> +::::::
>>> (I put it at the beginning).
>
>>Noooooo! It has to go at the END!!! That's where the nis maps are
>>inserted! You don't want them to override your own local entries!
>
> I have the line in the /etc/nsswitch.conf:
> passwd: files nis
It does no good there (well, not much). The lookup in files
(/etc/passwd) will already drag in the nis map that you have referenced
at the *head* of /etc/passwd (modulo provisos ..). You want just
"compat" (which is the default if you just blatt that file away, which
you should probably do ...). Well ... it's not clear from the manpage.
I leave you to judge:
Linux libc5 without NYS does not have the name service
switch but does allow the user some policy control. ...
Since most people only put a + at the end of /etc/passwd to
^^^^
include everything from NIS, the switch provides a faster
alternative for this case (`passwd: files nis') which doesn't
require the single + entry in /etc/passwd, /etc/group and
^^^^^^^^
/etc/shadow. If this is not sufficient, the NSS `compat' service
provides full +/- semantics. By default, the source is `nis',
but this may be overriden by specifying `nisplus' as source for
the pseudo-databases passwd_compat, group_compat and
shadow_compat. This pseudo-databases are only available in GNU C
Library.
In other words, what you have done is undefined behaviour. If you want
to lookup first in passwd and then in nis, you should EITHER have
compat as the mode in nsswitch and a +::::: at the END of passwd, OR
you should NOT have a +::::: line in passwd, but INSTEAD a "files nis"
line in nsswitch.
> So everything that is found in the file won't be read von a NIS map.
> But you might be right if the +:::::: entry renders the local file
> unread, so I will follow your advise.
>>You will have to use an automounter that can read sun automount maps.
>>
> Noooooo! The map entry in the client's auto.master file says:
> /net program:/etc/auto_net
"program:"? Eh what? That's not normal syntax. Linux auto.master files
look like
/- /etc/auto.bin -t 5
(that is a program entry). It's deemed to be a program if the
referenced file is executable.
> which means that no map file will be used from NIS
It means nothing of the sort. It simply says "look in the local file,
and if it's executable, run it insted of looking in it". What the
local executable does is up to it, and may well involve doing a ypcat
of some nis map.
> but the local one,
> and that it is an executable delivering necessary information for the
> automounter to do its work.
It's not in any syntax that matches any man page I own.
(auto.master(5))
The file has three fields separated by an arbitrary number
of blanks or tabs. Lines beginning with # are comments.
The first field is the mount point. Second field is the
map file to be consulted for this mount-point. The third
field is optional and can contain options to be applied to
all entries in the map. Options are cumulative, which is a
difference to the behavior of the SunOS automounter.
> The line:
> automount: files nis
> says explicitely "read the local map files first".
Uh, since when does your automount daemon pay any attention to
nsswitch? When the automount daemon is started, it is given its
arguments from lines in auto.master by the init script. Anyway,
that script is responsible for the interpretation of auto.master.
If you don't like its interpretation, rewrite the script. Its
job is to end up by starting
automount /- program /etc/auto.bin -t 5
(in the example I gave), and the init script will deduce that the map
type is "program" by noticing the x bit on the perms of auto.bin. Or at
least my startup script will! You can rewrite yours to do what you
please! If you figure out how to make it look in nsswitch.conf, tell
me.
> Since the map file
> for /net exists locally, it won't be taken from NIS and so it does not
> need to be understood by the Linux automounter.
Eh? All this is fantasy, unless you have some different automount
program than me! You may well have!
automount takes at least three arguments. Mandatory argu
ments include mount-point, map-type and map. Both manda
tory and optional arguments are described below...
> What you say is true though, for map files coming from NIS. The
> original entry in the Sun NIS server:
> /net -hosts
> could not be understood by the Linux server, if it was used.
>
> If I want to avoid a NIS map file to be used, I need only to put a
> local one in /etc.
Nope. You simply have to stop believing in magic, and instead provide a
causal chain to support your reasoning. I'm willing to believe you have
a new singing dancing automount program that is different to mine, but
you could provide evidence to back up your belief!
Peter
- Next message: Michael Heiming: "Re: LTSP"
- Previous message: Jean-David Beyer: "Re: large binary immediately SEGV's"
- In reply to: Rick Denoire: "Re: NIS and autofs and /net"
- Next in thread: Rick Denoire: "Re: NIS and autofs and /net"
- Reply: Rick Denoire: "Re: NIS and autofs and /net"
- Reply: Rick Denoire: "Re: NIS and autofs and /net"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|