Short question is: should we always do a
yum update --exclude=kernel-* to do an update in Fedora?
I was a bit surprised that when I got a new Fedora machine at work (it used to be Red Hat Enterprise Linux (RHEL)), the first time it started up, it asked me to do an update, and I naturally said yes, and for anything it asked, I just used the default answer (usually replying "yes" by pressing ENTER).
But it turned out that the kernel was somehow corrupted, having 3.6.9 and 3.6.10 components, and the machine would not boot up (would cause kernel panic), and had to use the second option to boot up in the boot menu. (the IT department told me it is the "last problem-free version", like a checkpoint version). But even that, the machine was very slow and my coworker told me later on that my kernel was running partly on earlier version and partly running with 3.6.9 or 3.6.10 components and they might not be totally compatible and that's probably why it was so slow.
My coworker knew about Fedora for quite a while, and he was able to fix it by doing a series of
yum remove and
yum downgrade, and making the kernel and header and component all back to an earlier version (I think it was 3.3.4 but I will have to go back to work and check).
So is it true that for Fedora, we always have to abort the default update request, and do a
yum update --exclude=kernel-*
to be safe that we won't get a kernel that is not yet stable or still in beta? This is somewhat counter-intuitive to me, as I know other update systems would usually use only the stable version, unless the user specifically types in a particular version which might be a beta version.
Is there actually a way to update only the stable components, or do we need to use
yum update --exclude=kernel-* all the time to be safe?
(this was a bit surprising to me, as anybody at work using Fedora can be affected by this, and it could be hundred or even thousands of people, so what is a more correct way of doing the system updates?)