I'm having difficulty locating the cause of a problem I've encountered in a version of the software I'm working on in which a yum update is being applied to update all packages on the system at the end of an installation wizard that prepares the initial image for the machine it's installed on.
There's nothing special about this update. We have a yum repository on a Raspberry Pi connected to the machine on the network, and this repository contains all the RPMs for the packages to update over the base image. A simple yum update -y is applied. This same update process currently works if I use the same Raspberry Pi with a different version of the software uploaded.
The problem occurs with one package, ntp, where after all the updates are applied, yum tries to cleanup the old version of the ntp package, and it just hangs... I've left it there for days and it never completes.
The RPM for the ntp package is the same one we use in every other version of the software, and this problem has never occurred in any other version. There is no difference in the repository itself, and the installer, as I said, simply calls yum update -y.
I checked using the command: sudo lsof -p $(pgrep yum) | head -10 as suggested in another StackExchange thread and the yum process that is running is looking at several DB files. It's pointless to worry about the running process though since the installation wizard is supposed to be automated and it was working before.
So I guess what I want to know is... since our installation wizard is massive and I have no leads to go on here, what are some things that could have changed that would cause the cleanup of the old ntp package to hang? I'm assuming SOMETHING must have changed in the installation process, but I can find no noticeable differences having to do with NTP or yum in general.
Any ideas? If it matters, this is Scientific Linux 6.4 with the 2.6.32-754 kernel.