I was using samba-1.9.18p5-1.i386.rpm and running along smoothly when I decided to update to samba-1.9.18p10-5.i386.rpm. After downloading the rpm from your site and upgrading (via "rpm -Uvh"), rpm segfaulted after the last hash mark was printed. Now rpm thinks both packages are installed: --- [root@threshold RPMS]# rpm -qa | grep samba samba-1.9.18p10-5 samba-1.9.18p5-1 --- and I can't uninstall either: --- [root@threshold RPMS]# rpm -e samba-1.9.18p10-5 Segmentation fault [root@threshold RPMS]# rpm -e samba-1.9.18p5-1 Segmentation fault [root@threshold RPMS]# rpm -e --test samba-1.9.18p5-1 Segmentation fault [root@threshold RPMS]# rpm -e --test samba-1.9.18p10-5 Segmentation fault --- And this is all after I rebuilt the database. I'm not a registered user, so I don't expect much help on this, but I thought you'd like to know. Thanks.
Sorry I didn't include this before: --- [root@threshold RPMS]# rpm -e -vv samba-1.9.18p10-5 D: counting packages to uninstall D: opening database in //var/lib/rpm/ D: found 1 packages to uninstall D: uninstalling record number 3593560 Segmentation fault [root@threshold RPMS]# rpm -e -vv samba-1.9.18p5-1 D: counting packages to uninstall D: opening database in //var/lib/rpm/ D: found 1 packages to uninstall D: uninstalling record number 4693592 Segmentation fault
do rpm -e --notriggers samba
Okay: --- [root@threshold RPMS]# rpm -e --notriggers samba-1.9.18p10-5 cannot remove /var/log/samba - directory not empty cannot remove /etc/codepages - directory not empty [root@threshold RPMS]# rpm -qa | grep samba samba-1.9.18p5-1 --- Ah, so it was a script that was segfaulting. Interesting.
Well, I was able to uninstall samba-1.9.18p5-1.i386.rpm using "rpm -e --notriggers" in combination with "rpm -q --scripts" and running the scripts by hand. I could then install samba-1.9.18p10-5.i386.rpm the same way. I don't know where the segfault came from, but it was probably actually some command in the scripts that had the error. Why doesn't rpm handle things like this gracefully? It would seem a trivial thing for rpm to tell me which part of the script had a problem... but this is a issue with rpm, not redhat per se. Anyway, thanks for the help.
It didn't handle it because of a bug in RPM, which I believe was fixed in a later version.