On Wed, 2006-05-03 at 11:33 -0700, Debian Bug Tracking System wrote:

> > On Sun, Apr 30, 2006 at 05:59:10PM +1200, Adam Warner wrote:
> > > I typically upgrade Debian for IA32 in a chroot environment of Debian
> > > for AMD64. nfs-common package upgrades consistently break (i.e. I've
> > > finally got around to reporting them!)
> > 
> > After discussing this with others (well, at least one), I've come to the
> > conclusion that this is not a bug -- you can't expect start-stop-daemon (nor
> > much else, really) to work properly without a mounted /proc, chroot or not.
> > Just mount /proc inside your chroot, and all should be well.

Steinar, many thanks for the tip! My fstab was and is correct according
to the Debian GNU/Linux AMD64 HOW-TO:
<http://alioth.debian.org/docman/view.php/30192/21/debian-amd64-howto.html>

This is my proc entry for my chroot environment:
proc            /ia32/proc      proc    defaults        0       0

I have tracked down the error to initscripts. This is
how /etc/init.d/mountall.sh parses fstab:

do_start() {
        #
        # Mount local file systems in /etc/fstab.
        #
        if [ "$VERBOSE" = no ]
        then
                log_action_begin_msg "Mounting local filesystems"
                mount -a -t proc >/dev/null 2>&1  # Ignore error message due to 
/proc already being mounted
                ES_TO_REPORT=$?
                mount -a -t noproc,nfs,nfs4,smbfs,cifs,ncp,ncpfs,coda,ocfs2,gfs
                ES=$?
                ES_TO_REPORT=$(($ES_TO_REPORT | $ES))
                if [ 0 = "$ES_TO_REPORT" ]
                then
                        log_action_end_msg 0
                else
                        log_action_end_msg 1 "code $ES_TO_REPORT"
                fi
        else
                log_action_msg "Will now mount local filesystems"
                mount -a -t proc >/dev/null 2>&1  # Ignore error message due to 
/proc already being mounted
                ES=$?
                [ 0 = "$ES" ] || log_failure_msg "Mounting proc filesystems 
failed with error code ${ES}."
                mount -a -v -t 
noproc,nfs,nfs4,smbfs,cifs,ncp,ncpfs,coda,ocfs2,gfs
                ES=$?
                if [ 0 = "$ES" ]
                then
                        log_success_msg "Done mounting local filesystems."
                else
                        log_failure_msg "Mounting local filesystems failed with 
error code ${ES}."
                fi
        fi

One can see that mountall.sh silently ignores all explicit fstab entries
to mount any proc filesystem (via the noproc directive)!

This is the most similar bug report:
"mountall.sh: additional proc mountpoints (for chroots) are not mounted"
<http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=359651>

Maybe this bug report could be merged with 359651?

Regards,
Adam



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to