#8077 NORM xs-0.4: Wrong ownership in /library/xs-rsync/ subdirs (Anaconda bug)
Zarro Boogs per Child
bugtracker at laptop.org
Thu Aug 21 05:40:44 EDT 2008
#8077: Wrong ownership in /library/xs-rsync/ subdirs (Anaconda bug)
-------------------------------+--------------------------------------------
Reporter: martin.langhoff | Owner: martin.langhoff
Type: defect | Status: new
Priority: normal | Milestone: xs-0.4
Component: school server | Version: not specified
Resolution: | Keywords:
Next_action: never set | Verified: 0
Blockedby: | Blocking:
-------------------------------+--------------------------------------------
Comment(by martin.langhoff):
Here is the log of my notes and diagnostics, as posted to fedora-devel-
list
Subject: livecd-creator / anaconda - strange problem with ownership
dropped
------------------------
From: Martin Langhoff <martin.langhoff at gmail.com>
Date: Thu, Aug 21, 2008 at 7:55 PM
To: Development discussions related to Fedora <fedora-devel-
list at redhat.com>
In summary: installing a LiveCD created livecd-creator, anaconda fails
to set the ownership & group of a couple of directories. Looking
inside the ext3 img (inside the squashfs), the directory owner/group
is set correctly and passwd and group files contain the correct
entries.
This is all on F7 - I am building a F7 livecd in a mock chroot running
F7. The package is a custom package, and it succeeds in installing
things correctly - no errors during the creation of the livecd, and
looking inside the squashfs/ext3 fs confirms that things are right.
The only "oddify" is that the directories that are showing this problem
are
/library/xs-rsync/pub
/library/xs-rsync/tmp
/library/xs-rsync/stage
and /library is a mountpoint (containing nothing else at the moment).
The rest of the system is on under a / (root) mountpoint.
Looking at /var/log/anaconda.log there are some odd entries saying
"INFO: set fc of /path to None" for several paths, including /library
and all its contents (literally, every file and directory in the
partition).
Any hints? Possible debug angles I should try?
BTW, if you want to see the %pre and %attrib creating the user/grp -
it is here http://dev.laptop.org/git?p=users/martin/xs-rsync.git;a=blob;f
=xs-rsync.spec.in;h=88a3f492ea8e3e878ea4fefb6d19c98f0f39ee88;hb=HEAD
- though I doubt it's interesting.
I'm off to grep through the anaconda sources - this is on
anaconda-11.2.0.66-1
cheers,
m
--
----------
From: Martin Langhoff <martin.langhoff at gmail.com>
Date: Thu, Aug 21, 2008 at 8:13 PM
To: Development discussions related to Fedora <fedora-devel-
list at redhat.com>
This is fairly strange - but I'm not sure if it carries the blame - we
see those INFO entries for each path in the non-/ mountpoints - /boot
and /library .
OTOH, I have a checkout of the matching anaconda code, and it's doing
isys.resetFileContext() which I followed all the way to lsetfilecon().
Will resetting the SELinux context reset the POSIX user/group too?
Ideas...?
----------
From: Martin Langhoff <martin.langhoff at gmail.com>
Date: Thu, Aug 21, 2008 at 9:28 PM
To: Development discussions related to Fedora <fedora-devel-
list at redhat.com>
On Thu, Aug 21, 2008 at 8:13 PM, Martin Langhoff
This was reported as #243479 -- and fixed in commit
4efdd691a6321d764acb0780d519e0ac625056ec which never got applied to
the F7 branch.
I have a backported patch (attached), but it's quite tricky to test
it. I have to rebuild the anaconda rpm (not an easy task!) to have the
rpm avaialble during the livecd creation. So it's not tested as it
stands.
----
--
Ticket URL: <http://dev.laptop.org/ticket/8077#comment:1>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system
More information about the Bugs
mailing list