FreeBSD 10.x + LiquidSoap + NFS == Server Hang

[Rick Macklem]

Kevin P. Neal wrote:
> On Fri, Jul 04, 2014 at 09:33:48AM -0700, Marc Fournier wrote:
> > I’ve made the change on two of my servers that have been exhibiting
> > NFS related issues, so we’ll see how this goes over the next
> > couple of days, but am curious as to when one would use ‘intr’ in
> > the first place, if by using it, it could cause NFS to hang like
> > this?  Granted, out of 11 servers using ‘intr’, this one server is
> > the only one that I’ve knowingly experienced the nfs_getpages, so
> > it seems to be something hard to trigger …
> 
> It used to be that if you had an nfs mount that was nointr, and the
> server
> died, then any client programs that were in the middle of an I/O
> would
> hang forever.
Well, until the server comes back up, actually.

There is no perfect solution, but I always used hard mounts (no intr, soft)
and just tried to make sure my NFS servers were reliable (same w.r.t. network
fabric), when I wore a sysadmin hat.

There is "umount -f", which can get rid of a hung server. Unfortunately, if
you do a "umount" (without -f) that is stuck on the crashed server's mount
point, "umount -f" won't work, because the other "umount" will have the mount
point locked up.

rick
ps: Both John Baldwin and Kostik have made changes to how certain signals are
    handled during syscalls (and also what happens when an nfs_getpages() fails)
    so the behaviour is dependent upon FreeBSD version.

> --
> Kevin P. Neal
>                                http://www.pobox.com/~kpn/
> "Oh, I've heard that paradox a couple of times, but there's something
> about a cat dying and I hate to think of such things."
>   - Dr. Donald Knuth speaking of Schrodinger's cat, December 8, 1999,
>   MIT
> _______________________________________________
> freebsd-stable at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-stable
> To unsubscribe, send any mail to
> "freebsd-stable-unsubscribe at freebsd.org"