> Date: Fri, 29 Mar 2013 17:07:14 -0400
> From: "Cliff McDiarmid" <[email protected]>
> To: "BLFS Support List" <[email protected]>
> Subject: Re: [blfs-support] Autofs problem on LFS7.2
>
        .
        .
> > I've been following this very long thread. Many thinks to akhiezer for 
> > the patience and support.
> > 
> > Cliff, it would be nice to write up a short summary of what was wrong 
> > and the ultimate fix.
>
> I can only echo what you have said Bruce.  Many thanks akhiezer for all your 
> help and patience!
>
> I will do my best to sum up this record breaking thread a.s.a.p. I still have 
> a few things to try.
>


Thanks - yr both welcome; 's no trouble, really.


(( It also had the by-product of refreshing and updating here on several areas 
of autofs; for quite a few years configs had just been 'set & forget' 
carried-forward. Two 'new' things in particular: how mount-paths and map-file 
permissions, are (now) handled.

For the former, I'd 'always' had the mount-path ready-made except for the last 
element (i.e. have /a/b/ ready-made), and had the mapfile contain only a single 
path-element (e.g. 'c' - i.e. no subdirs), such that automount was only 
creating/removing/maintaining the 'c' dir under '/a/b/' . This _may_ have been 
the behaviour in much-older autofs, or even the old amd (not the *PU company) 
automounter, &/or a workaround on some systems that didn't yet have 'mkdir -p' 
and the fallback alternative didn't work well, &/or simply misapprehension. So 
the statement that I made in one of the earliest posts in the thread, about an 
aspect of how automounter handles mount-point paths, is incorrect: the modern 
autofs can auto-gen the full mount-point path by itself; apols for the/any 
red-herring on that.

As for the map-file permissions: for ordinary file-type maps, I'd always used 
non-exec permissions, and just used the map pathname (and usually omitted any 
explicit 'file:' or 'file,sun:' unless really needed) in auto.master; and for 
anything else I'd use the likes of 'program:' explicitly in auto.master - i.e. 
for a program I'd basically never just use a pathname and then rely on the 
pathname having exec-permissions, for the path to be (auto-)interpreted as 
being 
a program rather than a simple 'file:'-type map. This last part is simply 
because I wasn't aware that automount would do such a behaviour. But, it 
(certainly now) does - the code's in the "if (st.st_mode & __S_IEXEC)" tests in 
the functions "read_file_source_instance()" and 
"lookup_name_file_source_instance()", of the src file 
"autofs-5.0.7/daemon/lookup.c"  .
))


rgds,
akh





--
-- 
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to