On Mon, Oct 30, 2006, martin f krafft wrote:
> This is not posix, it cannot be empty. I suggest ':;;', which is
> valid.
Thanks; I knew about it for if/then/else, and didn't think it would
apply to other constructs as well.
--
Loïc Minier <[EMAIL PROTECTED]>
also sprach Loïc Minier <[EMAIL PROTECTED]> [2006.10.30.0050 +0100]:
> + *)
> +;;
This is not posix, it cannot be empty. I suggest ':;;', which is
valid.
--
.''`. martin f. krafft <[EMAIL PROTECTED]>
: :' : proud Debian developer, author, administrator, and user
`. `'`
Hi,
With the input of Martin, and following my own suggestions:
On Sat, Oct 28, 2006, Loïc Minier wrote:
> On the topic of mounts, I'd like to add:
> * a sanity check to popup a shell if a mount point isn't an empty
>directory after umount
> * reverse order of umounts so that /pro
On Sun, Oct 29, 2006, martin f krafft wrote:
> First, I would use a case statement instead of echo | grep.
Right.
> Second, how about:
> local UMOUNT_OUTPUT
> if UMOUNT_OUTPUT="$(LC_ALL=C umount "$BUILDPLACE/$1")"; then
> # it worked
> else
> # it did not
> fi
Ah, thanks! For some reason,
also sprach Loïc Minier <[EMAIL PROTECTED]> [2006.10.28.1818 +0200]:
> Could you please try the attached patch? I've tested it in both
> working and not working cases, and it seems do what I want it to
> do: ignore only your particular type of umount error.
Yup, seems to work.
> +local UM
On the topic of mounts, I'd like to add:
* a sanity check to popup a shell if a mount point isn't an empty
directory after umount
* reverse order of umounts so that /proc is umounted last
* perhaps a sanity check to grep /proc/mounts for the base path to the
chroot and see if anything mou
tags 391390 + patch
stop
Hi Martin,
On Fri, Oct 06, 2006, martin f krafft wrote:
> -> unmounting proc filesystem
> umount: /proc: device is busy
> umount: /var/cache/pbuilder/build/27006/proc: not mounted
> umount: /proc: device is busy
> umount: /var/cache/pbuilder/build/27006/proc: not
also sprach Junichi Uekawa <[EMAIL PROTECTED]> [2006.10.09.0702 +0200]:
> > diamond:~> PID=14692
> > #[130,303]
> > diamond:~> grep -q "pbuilder/build/$PID/proc" /proc/mounts && echo /proc is
> > mounted under pbuilder chroot $PID
Hi,
> > It is difficult to distinguish between not being able to umount /proc,
> > and /proc not being mounted in the first place, and the bet is done on
> > the safe side of things, since after all those checks have passed,
> > we're going to 'rm -rf' the whole tree, and it is possible for a
> >
also sprach Junichi Uekawa <[EMAIL PROTECTED]> [2006.10.08.1204 +0200]:
> It is difficult to distinguish between not being able to umount /proc,
> and /proc not being mounted in the first place, and the bet is done on
> the safe side of things, since after all those checks have passed,
> we're goin
Hi,
> > I don't think pbuilder should really care about user errors. Shooting
> > your foot, that's fine. You can even remount /proc to recover from
> > this state.
>
> How does umounting /proc mean shooting myself in the foot? How is it
> an error?
The check wasn't added for no reason.
It is
Hi,
> For testing purposes, I had previously umounted /proc already.
> I don't think pbuilder should try to umount it if it's not mounted.
Hmm...
I don't think pbuilder should really care about user errors. Shooting
your foot, that's fine. You can even remount /proc to recover from
this state.
also sprach Junichi Uekawa <[EMAIL PROTECTED]> [2006.10.06.1506 +0200]:
> I don't think pbuilder should really care about user errors. Shooting
> your foot, that's fine. You can even remount /proc to recover from
> this state.
How does umounting /proc mean shooting myself in the foot? How is it
a
Package: pbuilder
Version: 0.159
Severity: normal
-> unmounting proc filesystem
umount: /proc: device is busy
umount: /var/cache/pbuilder/build/27006/proc: not mounted
umount: /proc: device is busy
umount: /var/cache/pbuilder/build/27006/proc: not mounted
W: Retrying to unmount proc
umount: /var/
14 matches
Mail list logo
