Fixing the network stop script

DJ Lucas dj at lucasit.com
Wed May 28 19:34:06 PDT 2003


Hola guys and gals.  I've actually been working on the problem with 
mounted network filesystems and dhcp clients.  ie: the nework being 
killed by sendsignals before the filesystems are unmounted.  I wanted to 
pick a few people's brains, and have a look at the following to see if 
there is anything that I might have missed, not considered, consequences 
etc...

Consider this:  Move network to K45network (before sendsignals) so the 
network is down before sendsignals runs.  This has been suggested time 
and time before, but that brings bigger problems. Network filesystems 
are not unmounted before the network goes down...unclean fs's, file 
locks, long timeouts, etc...  Now we add a netfs script at K43netfs that 
looks at the /etc/mtab file and grabs all instances of nfs, smbfs, and 
later, others (ncpfs??) as they go into the book.  Take that list and do:

fuser -km $MY_LIST_OF_MOUNTED_NETWORK_VOLUMES
umount $MY_LIST_OF_MOUNTED_NETWORK_VOLUMES

This sounds pretty destructive, and it is, but no more destructive than 
the sendsignals script we currently use.  It works (by hand) in my 
minimal test cases just fine.  The testcase I used was simply cd'ing 
into a directory on a mounted nfs volume, and firing up vi.  It kills 
the console and vi and brings up a login screen.  It _seems_ perfectly 
fine for me in RL0 and RL6.  Anyone think that this is _too_ destructive 
in RL1 and RL2?  How about any problems or different situations that I 
might have missed?

Thanks in advance,

DJ

-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe blfs-dev' in the subject header of the message



More information about the blfs-dev mailing list