Re: util-linux: bug #19039 still true ?
On Tue, Feb 01, 2000 at 08:04:07PM +0100, Thierry Laronde wrote:
> Can someone else have a look to the bug #19039.
Hmm, I think the information in your mail is enough :)
> This old bug has been reassigned from sysvinit to mount, because of the
> following :
>
> "If any remote filesystems are mounted, the system will hang after a shutdown
> -r and refuse to properly unmount even the local filesystems if any remote
> computer is not responding. For instance, if I mount a partition using NFS
> from machine alex, and then alex goes away, my machine will refuse to shut
> down because it apparently is trying to unmount from a machine that is no
> longer turned on (or whatever). This can cause corruption on LOCAL
> filesystems too."
Not true - at least not anymore. But I think even 2.0.x shut down correctly
after a (long) timeout. This happens here:
pulsar:~# /usr/bin/time umount -a -t nfs
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Timed out
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Timed out
0.00user 0.03system 2:00.35elapsed 0%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (129major+20minor)pagefaults 0swaps
I cut off the network to test this. Kicking portmap on the server, shutting
down the network etc. did not do more harm either.
> and Miquel van Smoorenburg reassigning it wrote:
>
> "As you can see, I reassigned this bug to "mount". Note that this is actually
> a kernel problem I am afraid as what you need to fix this is "hard
> unmounting", something not implemented in the Linux kernel.
>
> But a workaround would be a timeout in the mount program for NFS mounts.."
Wrong. The Kernel blocks the program. mount can't do anything about it.
But the kernel implements the timeout now. Also the current mount has
a force option (supported since some 2.1.x kernel).
> If this not solved (at least by the use of the `intr' option when mounting
> a NFS) ?
It is solved. Close it.
cu
Torsten
--
Torsten Landschoff Bluehorn@IRC <torsten@debian.org>
Debian Developer and Quality Assurance Committee Member
Reply to: