Bug 1601011 - file /usr/bin/ldd conflicts between attempted installs of glibc-2.27.9000-39.fc29.i686 and glibc-2.27.9000-39.fc29.x86_64
Summary: file /usr/bin/ldd conflicts between attempted installs of glibc-2.27.9000-39....
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: glibc
Version: rawhide
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Carlos O'Donell
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2018-07-13 14:56 UTC by Yanko Kaneti
Modified: 2018-07-13 21:08 UTC (History)
13 users (show)

Fixed In Version: glibc-2.27.9000-40.fc29
Clone Of:
Environment:
Last Closed: 2018-07-13 20:44:29 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Yanko Kaneti 2018-07-13 14:56:25 UTC
Description of problem:
Trying to update glibc to glibc-2.27.9000-39.fc29
.....
	file /usr/bin/ldd conflicts between attempted installs of glibc-2.27.9000-39.fc29.i686 and glibc-2.27.9000-39.fc29.x86_64

Comment 1 Florian Weimer 2018-07-13 14:57:56 UTC
Please provide the complete DNF error message.

Comment 2 Yanko Kaneti 2018-07-13 15:00:00 UTC
Its not dnf but rpm -Fhv

Comment 3 Florian Weimer 2018-07-13 15:08:58 UTC
Oh, I see.

Carlos, your filelist reorganization was incorrect.  Did you plan to make this change?  You said that

    The changes were tested by carrying out a file-by-file comparison
    to ensure no file changed packages or installed location after
    the patch.

but clearly this wasn't the case.

Comment 4 Carlos O'Donell 2018-07-13 16:59:55 UTC
(In reply to Florian Weimer from comment #3)
> Oh, I see.
> 
> Carlos, your filelist reorganization was incorrect.  Did you plan to make
> this change?  You said that
> 
>     The changes were tested by carrying out a file-by-file comparison
>     to ensure no file changed packages or installed location after
>     the patch.
> 
> but clearly this wasn't the case.

I'm reviewing this right now.

Comment 5 Carlos O'Donell 2018-07-13 17:39:56 UTC
(In reply to Carlos O'Donell from comment #4)
> (In reply to Florian Weimer from comment #3)
> > Oh, I see.
> > 
> > Carlos, your filelist reorganization was incorrect.  Did you plan to make
> > this change?  You said that
> > 
> >     The changes were tested by carrying out a file-by-file comparison
> >     to ensure no file changed packages or installed location after
> >     the patch.
> > 
> > but clearly this wasn't the case.
> 
> I'm reviewing this right now.

So, something clearly went wrong, the final rawhide builds shows a file list difference, that is to say my expand-rpms.sh / compare-rpms.sh scripts show a difference. If I'd seen it I wouldn't have pushed a rawhide build.

I had what I thought was a minor merge cleanup before I pushed my commits, but clearly I made a mistake somewhere. Sorry for that. I'll fix this up right now.

Comment 6 Adam Williamson (Red Hat non-Fedora) 2018-07-13 19:19:22 UTC
For the record, this broke today's Rawhide compose as the sudden embiggening of the glibc package caused the ARM minimal disk image package set to be too big for the root filesystem size. https://pagure.io/dusty/failed-composes/issue/532

Comment 7 Carlos O'Donell 2018-07-13 19:31:07 UTC
I had *just* verified everything when I did my git pull --rebase, and ran into a merge conflict.

It's a one line fix.

-%files -f rpm.filelist
+%files -f glibc.filelist

Florian's changes to add POWER9 were just at the right place to cause a merge conflict, and I picked Florian's changes without noticing the rpm.filelist->glibc.filelist reversion. The result is that glibc.rpm has *all* the files from all the packages.

Teaches me that I should always just change the names of data files if they change their semantic meaning in the build process. After the reorganization the rpm.filelist is the "master" file list but shouldn't get used for any package file lists. So I've renamed 'rpm.filelist' to 'master.filelist' to denote the semantic change and avoid any confusion between earlier versions of the spec file and the current version.

This will be fixed in glibc-2.27.9000-40.fc29. Doing a scratch build again just out of paranoia.

Sorry for the breakage.

Comment 8 Carlos O'Donell 2018-07-13 20:44:29 UTC
New build in progress:
https://koji.fedoraproject.org/koji/taskinfo?taskID=28216160

Comment 9 Carlos O'Donell 2018-07-13 20:45:02 UTC
Scratch build and local mock build all passed comparison tests again.


Note You need to log in before you can comment on or make changes to this bug.