yum update excluding kernel ... why?

[madcat]

Hello everyone. I've recently been assigned admin duties for servers running Plesk on CentOS. Both Plesk and the OS were outdated, so I have run 'yum update' to upgrade to CentOS 4.5 and used the Plesk updater to update to Plesk 8.1.1 and overall it was a fairly smooth process.

I've noticed that /etc/yum.conf contains this line:

exclude=kernel* sendmail*

I understand the "sendmail*" entry, since Plesk uses Qmail. However, I do not understand why the kernel is excluded. One server is running the kernel "2.6.9-42.0.10.ELsmp", others are running "2.6.9-42.0.3.ELsmp", but currently, the latest kernel is "2.6.9-55.0.2.EL".

Does anybody know of a good reason why I can't upgrade to that kernel? Does the Plesk install software add that line to yum.conf, or would it be something that the previous admins of these servers did? Has anybody else already upgraded to "2.6.9-55.0.2.EL"? If so, did you encounter any problems?

Thanks.

 

[breun]

Plesk didn't do that. Some providers/admins do this, because they don't trust kernel upgrades. A kind of "if it ain't broke, don't fix it" idea probably. The problem is of course that you do need to know if there aren't any security issues with the kernel you're running that might impact your situation.

We run the latest CentOS kernel on all of our servers, works great. And the Plesk installer should have removed sendmail (Plesk uses qmail). Excluding it in the yum config was probably 'just making sure', but you can safely take that exclude out as well.

 

[wjtech]

Some data centers offer a kernel optimized for their network and boxes (e.g., DedicatedNow), and make provisions for upgrading using their scripts. Check with your provider.

In any event, it's a good idea to exclude the kernel and certain other key packages (mysql, php, etc.) where an unintended upgrade might hammer Plesk. Of course, you must stay aware of when security updates are needed. Then, just unexclude (?) the package and update/upgrade.

John

 

[atomicturtle]

Ive yet to see a hosting company "optimized" kernel that really was. They generally do it so they can have one multi-distro kernel. The other conditions I can only chalk up to someone being OCD.

Provided you're using "update" rather than install (or rpm at the command line) a kernel update is going to be added to the system, and will not replace the currently running kernel. You can set a new kernel to boot once from inside of grub (very handy for testing) with:

# grub
grub> savedefault --default=X --once

Where X is the position of the kernel in /etc/grub.conf, starting with 0. (Position 1 -> 0, Position 2 -> 1, Position 3 -> 2...)

 

[breun]

GRUB's fallback option is also really nice: http://www.gnu.org/software/grub/manual/html_node/Booting-fallback-systems.html