Red Hat Bugzilla – Bug 112792
Installer fails with rpmdb: fatal region detected message
Last modified: 2007-11-30 17:10:34 EST
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
Description of problem:
Got an rpmdb error while installing FC 1 in the graphical mode.
Somewhere in the middle of the installation I switched to F1 screen
and saw tons of "rpmdb: fatal region detected; run recovery"
messages. It does not really matter what to select to install, but
rather the number of packages to install: installing the very minimal
configuration does not seem to produce such an error.
After the installation is finished I can not boot into Linux.
I have tried this several times with no success.
Looks like it installs some number of first packages and than does
not install the rest because of this error.
I have installed RH 8 on this machin with no problems. But both RH9
and FC1 fails with the identical errors.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Start the installation process on my computer
2. Switch to F1 screen and observe....
This problem is very well reproducable on my computer. The CDs are
fine, I checked them. The HDD is ok too.
I realize there is not much information in this posting to try to
isolate the problem, but if you tell me what information is needed I
will provide it.
So the CDs have been checked with the built-in mediacheck
functionality? If so, can you boot with 'linux allowcddma' and see if
I tried the linux allowcddma boot option, but same failure occurred.
I don't know whether the same file was causing the problem or not.
Hope this helps.
I also get the same error in an NFS RH9 installation as well. Haven't
tried any earlier versions as I don't have media here.
I get the same errors.
When looking at anaconda's install.log file, it reports:
"Error: db4 error(-30978) from db->curso: DB_RUNRECOVERY: Fatal error,
run database recovery"
This is on both Core 1 and Core 2 test.
I do not get this error message on RH9's anaconda.
Oh... and I've checked RAM with several passes of memtest86 to no avail.
This should be better in newer releases.