Hello again-

as I mentioned in my debut thread here, I am a mostly Windows admin who recently inherited management of a RedHat 5.x server. One of the unfortunate discoveries I have made is that the box has not been updated in a long, long time. There is an active subscription with RedHat, and I've logged in to the customer portal and there I can see the long list of updates and bug fixes available for install.

Should I start with the oldest and move forward?

Are there any updates that are going to require a reboot? I don't see that in the descriptions of any of the updates I've looked at, but seeing as how there are over 400 (I know, it's bad), I have not read them all. I want to be sure that I know when/if a reboot is required and that I am in control of that process.

You may be wondering why I don't just call RedHat up and ask them these questions. Well, the server was purchased from Dell with RedHat pre-installed, and apparently when you buy a server configured thusly, you must go to Dell for support. They may very well have great RedHat support, but after waiting on the phone for 20 minutes for a RedHat support guy to tell me 'Sorry, you have to call Dell', I didn't want to get on the phone again.

thanks

The two packages requiring a reboot : glibc , kernel.

So whenever it is convenient, you can do # yum install glibc kernel
Then the packages will be updated, and you can reboot.

The other "400" updates : No reboot is required.


.

I would ad that for the other 400, you may have to restart some of the services they apply to in order to get the updated version running.

Seriously:
1. start with (ideally 2) full backups.

2. get spare PC and install the backup.
Ideally, you'd want the same basic HW (to check kernel+drivers) and just enough disk space to do basic app tests

3. you can change the RH subscription to point to the new PC.
Subscriptions are not locked to a machine, just the total number of machines registered at any one time

4. install all updates on new PC
5. test

6. when all good, re-subscribe live prod, do backups again (it will have chnaged0 and update prod

7. you should find out how many subscriptions you have and ensure you have at least one for each prod and at least one for a single dev/test box.
Actually, you can cheat a bit by using Centos on some/all of the boxes.
Its a free rebuild of RHEL, includes updates, but no support.

As you can see, you could skip some of those steps, but its high risk.
If this is a serious server, then doing it properly is the cost of doing business.
I'm sure you have the equivalent steps for your MS servers.

Can't remember if I've pointed you at this before, so http://www.linuxtopia.org/online_boo...ion/index.html

1 members found this post helpful.