Caveats when upgrading from RHEL 5.4 to RHEL 6.3 on hundreds of production servers [duplicate]

Question

Possible Duplicate:
Caveats when upgrading from Java 6 to Java 7 on hundreds of production servers

I am about to upgrade the RHEL version from RHEL 5.4 to RHEL 6.3 in a few hundreds of production servers.

The upgrade will be tested on a staging environment on virtual machines before production deployment.

Are there any backward-compatibility issues with RHEL upgrades?

Are you performing in-place upgrades? If so, please see: Why is it so difficult to upgrade between major versions of Red Hat and CentOS? — ewwhite, Jan 30, 2013 at 14:01
It's not in-place. The server is re-installed, and the existing software packages - in-house and external open source tools - are installed on the new operating system, so I guess my main concern is software package compatibility. — Adam Matan, Jan 30, 2013 at 14:10
@AdamMatan This is your environment, so you'll have to test the software side. I've given some general advice about basic differences, though. — ewwhite, Jan 30, 2013 at 14:17

Community · Accepted Answer · 2017-04-13 12:14:13Z

I wouldn't run the upgrade process this way if you're planning to perform in-place transitions from EL5 to EL6. It's not an easy or clean process.

However, you have the right idea of testing this in a limited-capacity first.

Regardless of deployment technique, the biggest issues I saw in moving from EL5 to EL6 were:

Dealing with deprecated packages
Needing a much larger /boot partition (from 100Mb to 300Mb)
A large amount of performance tuning needed to deal with the newer scheduling algorithms, different kernel and sysctl defaults, etc.
Learning about the tuned-adm framework. Also see: Openfiler iSCSI performance

1 Answer 1