On Thu, Jul 03, 2003 at 01:59:04AM -0400, Edwin Robertson spoke thusly: >Using rpm to upgrade your running kernel is a really bad idea no matter >what version of Red Hat, especially if you are using scsi. Basically what
Yes, SCSI. That's the only difference between my current desktops and the server boxes I am using. You are probably right there. >it does is rip the modules for the running kernel right out and install >the new kernel modules. In the case of scsi, you can write the initrd >because you have no loop back devices once you've ripped out the modules. OK. >Unless of course you have compiled your own kernel and compiled it all >directly in, which in that case I would doubt you would want to be >upgrading to one of Red Hat's kernel rpms. What you should do instead of >-F or -U, is just a regular -I install. This will install the two kernels >along side each other and once you are absolutely sure you have no >problems with the new kernel then you can remove the old kernel with rpm >-e kernel-<version>, make sure to include your version :). I used to compile kernels individually for all my servers, but that became an administrative && maintenance nightmare. So now I only do it for specialized hosts (routers/firewalls etc) which is why I'm just getting into the RPM thingy :) And yes, I found out that doing -Fvh keeps no backup kernel in place also (the hard way). I found this via a quick websearch : https://rhn.redhat.com/errata/RHSA-2003-025.html http://www.redhat.com/support/resources/howto/kernel-upgrade/ I'll go through that, but yes -- looks like I've been jumping in fire all this time :-) Oh well, time to go beat my head against the wall and chant 'daft luser, daft luser' repeatedly. Thanks alot for the advice Erwin & Brady. -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list