[nycbug-talk] nullfs, jails and quotas
Ray Lai
nycbug
Thu Jan 12 14:19:28 EST 2006
On Thu, Jan 12, 2006 at 02:14:22PM -0500, Okan Demirmen wrote:
> On Thu 2006.01.12 at 14:09 -0500, Isaac Levy wrote:
> > Hey Charles, All,
> >
> > Not a very timely response, but I feel I should weigh in on this issue.
> >
> > On Jan 9, 2006, at 7:17 PM, Charles Sprickman wrote:
> >
> > >Here's a quick tip:
> > >
> > >Don't try using nullfs to mount /usr/ports (or anything else) into
> > >a jail and then enable quotas on the jail's partition. Really bad
> > >juju there.
> > >
> > >This is on 6-stable. On running quotacheck after killing off all
> > >jails, that process was hung, and anything else that touched that
> > >partition went into a "disk wait" state that was unkillable.
> > >Running a shutdown just left the machine hanging in a not-quite-
> > >dead state. It seems like the kernel will block forever on any
> > >processes waiting for that partition. A reboot alone did not clear
> > >it, the background fsck also hung.
> > >
> > >Just a friendly warning for anyone doing that kind of tinkering
> > >remotely - always unmount your nullfs mounts before screwing with
> > >quotas...
> >
> > Nullfs is outright unsuported, as I've said during the course of
> > several lectures on jail(8). The man page for mount_nullfs(8) states:
> >
> > <snip>
> > BUGS
> > THIS FILE SYSTEM TYPE IS NOT YET FULLY SUPPORTED (READ: IT
> > DOESN'T WORK)
> > AND USING IT MAY, IN FACT, DESTROY DATA ON YOUR SYSTEM. USE AT
> > YOUR OWN
> > RISK. BEWARE OF DOG. SLIPPERY WHEN WET.
> >
> > This code also needs an owner in order to be less dangerous -
> > serious
> > hackers can apply by sending mail to <hackers at FreeBSD.org> and
> > announcing
> > their intent to take it over.
> > <snip>
> >
> > Charles: I'm dreadfully empathetic if this cost you time and energy.
>
> and as ray _almost_ eluded to a bit earlier, there's a reason why the
> layered filesystems were pulled from at least openbsd.
I should have been more explicit instead of assuming everyone was
in the know. =)
-Ray-
More information about the talk
mailing list