Red Hat Bugzilla – Bug 109734
Install fails on bad package, when it should continue
Last modified: 2007-11-30 17:10:33 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5)
Description of problem:
I was upgrading my RH 9 install with Core 1. I was on disk 3. The
installer reported some problem with the kernel source package and
said it couldn't continue.
Aaaaaargh! You can't abort once you have changed the disk, not on a
package like that. I could see aborting on really important packages
like the kernel or really important libs such as glibc. In fact, those
should be carefully checked before you even try to upgrade. The kernel
source is a failure the user can recover from easily.
What should have happened is it should have reported the error, tossed
any packages that were dependant on it, and continued on with the
rest. It was almost done.
A newbie user would be wringing his hands right now, wondering whether
his machine was hosed. (Thankfully, I tried it on a machine I could
afford to hose.)
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install or upgrade
2. Have a package that fails due to bad media or some other corruption
Actual Results: Installer aborted with a message regarding a
non-critical package, and rebooted
Expected Results: Installer should have continued, leaving out bad
package and those packages dependant on it.
In case it's a real corruption in the iso's, failure from the log:
error: unpacking of archive failed on file
0cf1b: cpio: MD5 sum mismatch
I think that if there are no dependencies on a package that is
considered bad, Anaconda should continue. If there are dependencies,
see if there is anything that is mandatory for system operation that
will be broken and proceed as determined.
*** This bug has been marked as a duplicate of 68376 ***
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.