> Obviously that's not the real fix, since it precludes names with quoted
> spaces.  The bug is presumably in auto.net, since it shouldn't be
> generating '\ ' if it really wants a continuation.

I know, I just wanted to test to see if it worked ;) The main reason for 
pointing it out is that sun's automounter will work properly in those cases. 
Here's what the map looks like above, including and below the home directory 
section.

gridengine      rogueleader:/export/gridengine
nfshomes        \
/brinda   deimos:/export/nfshomes2/brinda \
/dtsouma   deimos:/export/nfshomes2/dtsouma \
/ijdar   deimos:/export/nfshomes2/ijdar \ 
...
/gvijay   deimos:/export/nfshomes2/gvijay \ 
/gcraigm   deimos:/export/nfshomes1/gcraigm \ 
/weszka   deimos:/export/nfshomes1/weszka \ 
/pooja   deimos:/export/nfshomes1/pooja \ 
/msubotin   deimos:/export/nfshomes2/msubotin
volumes         deimos:/export/projects/archive/volumes
volumes2        deimos:/export/projects/archive/volumes2


> 
> Not really.  autofs only sees lookups at its mountpoint level (ie, /fs2
> in your case).  I guess if nfshomes isn't mounted from the NFS server,
> it could also trigger at that level, but the common case where the
> exports are dir and dir/subdir, once you mount dir, it can't trigger on
> any lookups under there (that's effectively a direct mount).
> 
> Perhaps you could modify your auto.net and your autofs setup, so that
> nfshomes is the autofs mountpoint, and each home dir is treated as a
> separate key?
>

Is there a way to embed a map a level or two deep? In our case the entire /fs 
tree has automount items in it, not just /fs/nfshomes, so starting autofs on 
/fs/nfshomes isn't possible.

Unfortunately, due to the 4095 character/line limit in the autofs and sun 
automounter code we're probably just going to set it up as a seperate mount 
tree (/nfshomes or something). Trying to scale out more than a few dozen home 
directories is presenting a large headache. The only thing that we've found to 
do arbitrary layering is amd, but training users not to use /a/ or /.automount 
is futile and it has some stability issues on solaris + dfs.


Thanks,
 -Mike
 


_______________________________________________
autofs mailing list
[EMAIL PROTECTED]
http://linux.kernel.org/mailman/listinfo/autofs

Reply via email to