Bug 504308

Summary: Anaconda release upgrade process leave OS crippled over trivial dependency issue
Product: [Fedora] Fedora Reporter: Jonathan Cervidae <jonathan.cervidae>
Component: anacondaAssignee: Anaconda Maintenance Team <anaconda-maint-list>
Status: CLOSED DUPLICATE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: urgent Docs Contact:
Priority: low    
Version: 11CC: anaconda-maint-list, jonathan.cervidae, rmaximo, vanmeeuwen+fedora
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2009-06-05 14:42:11 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Jonathan Cervidae 2009-06-05 14:11:16 UTC
Description of problem:


Version-Release number of selected component (if applicable):

[root@jaydee squashfs]# ./usr/lib/anaconda-runtime/loader/init --help
(running in test mode).

Greetings.
anaconda installer init version 11.5.0.59 starting


How reproducible:

Have a dependency conflict that anaconda cannot resolve.

Steps to Reproduce:

Obvious from how reproducible.
  
Actual results:

Anaconda exits with error code 1, leaving my system completely crippled. It took me 8 hours to place it back together again manually with yum. The actual component with a dependency issue was just something to do xsane. Anaconda decided to exit and half my packages were now fc10, half fc11 and many were simply missing. As you can imagine it was painful to fix by hand by yum in such a state.

Expected results:

Anaconda not to exit because it cannot resolve a dependency problem. At the very minimal level it should offer the user the chance to just delete the problem component, which is what I would have done since it was just something to do with my scanner I could have fixed later and then continued as normal. Exploding over something so minor and causing such incredible damage to my OS is not what I expected from the recommended method of upgrading one Fedora release to another. It would have been easier for me to do it by hand with yum.

Additional info:

This should be moved right to the top of your list. Of course everyone says that their bug is the most important but this one destroyed my OS over something that didn't even matter and you can easily fix it just by offering to delete the problem component if you don't want to do anything more clever for now. Imagine if this happened to someone who didn't know how to use lvm, cryptsetup, rpmdb etc who had this problem (i.e. someone like me 12 hours ago) but that they were also a non-technical user who couldn't poke about at it to work out what went wrong. You would reasonably expect them to never use Fedora again. I'm going to assume this problem exists in all Anaconda versions and is not related to FC 11 being new that's so very serious I can't believe it did that!

Please don't be offended by the format of the bug report either, I appreciate the free use of Fedora but was horrified that something of this nature exists in it and felt it important to convey that horror so it might get the right priority.

Comment 1 Chris Lumens 2009-06-05 14:42:11 UTC
The problem here is that you're doing a split media upgrade, and each individual disc is its own transaction.  We can determine conflicts within transactions, but not between different transactions.  For various reasons, the fix here really needs to be that we merge all the discs into one big transaction so we can detect these and other errors.  Luckily we already have a bug to track this issue.

Aside from that, package conflicts are packaging bugs and need to be fixed in the packages themselves.  There are simply too many situations where package conflicts result in the inability of the user (let alone a program) to make any correct decision to solve this via the UI.

*** This bug has been marked as a duplicate of bug 492629 ***