revisor

#216 Failed Live-Spin results in left over bind mounts
Closed: Fixed Opened by jsteffan.

Closed: Fixed
Reopen Issue

Traceback (most recent call last):
File "./revisor.py", line 201, in
revisorBase.run()
File "/home/jon/Fedora/Revisor/revisor/revisor/cli.py", line 48, in run
self.base.lift_off()
File "/home/jon/Fedora/Revisor/revisor/revisor/base.py", line 532, in lift_off
self.buildLiveMedia()
File "/home/jon/Fedora/Revisor/revisor/revisor/base.py", line 904, in buildLiveMedia
if not liveImage.setup(livemedia_targetsize, build_dir="/var/tmp/revisor-livecd", yum_installroot=self.cfg.yumobj.conf.installroot, yum_cachedir=self.cfg.yumobj.conf.cachedir):
File "/home/jon/Fedora/Revisor/revisor/revisor/pilgrim.py", line 448, in setup
os.symlink("../proc/mounts", self.build_dir + "/install_root/etc/mtab")
OSError: [Errno 17] File exists

Our startup/preclean code does not deal with this, yet.

with what? unmounting the loopmounts? or with loopmounts that are build upon existing os.img/etc files? Thing is, this file should only exist inside the %s/data/os.img filesystem, not on 'the outside' or host filesystem.

with what? unmounting the loopmounts? or with loopmounts that are build upon existing os.img/etc files? Thing is, this file should only exist inside the %s/data/os.img filesystem, not on 'the outside' or host filesystem.

It's the bind mounts into the chroot when doing a live spin.

It's the bind mounts into the chroot when doing a live spin.

There's an os.img which gets loopmounted, then mkdir() statements for /etc, /usr, /bin and /sbin, then a symlink() for install_root/proc/mounts (/proc has been bind mounted) to install_root/etc/mtab (of which the /etc/ directory has just been created). IMO, if install_root/etc/mtab already exists, it should be in the os.img, shouldn't it? If bindmount install_root/proc/mounts already exists, it should matter since it's the source for the link and should already exist, right?

There's an os.img which gets loopmounted, then mkdir() statements for /etc, /usr, /bin and /sbin, then a symlink() for install_root/proc/mounts (/proc has been bind mounted) to install_root/etc/mtab (of which the /etc/ directory has just been created). IMO, if install_root/etc/mtab already exists, it should be in the os.img, shouldn't it? If bindmount install_root/proc/mounts already exists, it should matter since it's the source for the link and should already exist, right?

BTW, if os.img doesn't get deleted that still indicates we're not unmounting properly. I think though that is the losetup -a loopmount

BTW, if os.img doesn't get deleted that still indicates we're not unmounting properly. I think though that is the losetup -a loopmount

I'm thinking this has to do with the lack of a losetup loopmount deletion, keeping os.img from being deleted.

I'm thinking this has to do with the lack of a losetup loopmount deletion, keeping os.img from being deleted.

http://git.fedoraproject.org/?p=hosted/revisor;a=commitdiff;h=2a5b2a27b8277bd40dd9132918bf8f5c7dc6a7db

http://git.fedoraproject.org/?p=hosted/revisor;a=commitdiff;h=2a5b2a27b8277bd40dd9132918bf8f5c7dc6a7db

Metadata Update from @jsteffan:
- Issue assigned to kanarip
- Issue set to the milestone: 2.0.4 Release
Metadata
None
None
None
None
Static archive. pagure.io was sunset in 2026; this is a read-only snapshot.
Red Hat Logo

Fedora is sponsored by Red Hat. Learn more about the relationship between Red Hat and Fedora »

© Red Hat, Inc. and others.