Skip to main content
Ask Question
phale
Systems Administrator IV
Asked a question 7 months ago

I'm currently running my cluster on RHEL 6.10, which will be going EOL on November 31, 2020. I had put in a support ticket back in 2018 asking if we could use the redhat-upgrade-tool to upgrade the head node and compute images to RHEL 7.x. The answer back then was "not recommended" and to install a second head node into the cluster with a newer OS, transfer services to it, and then rebuild the compute images. Just wanted to see if this is still the case or if someone has successfully upgraded the OS using the redhat-upgrade-tool.

Where am I?

In Bright Computing, Inc. you can ask and answer questions and share your experience with others!

That is still the best course of action.  There can be  "gotchas" in an upgrade from RHEL 6 to 7 that you may want to workout offline before upgrading your cluster like device driver or software package issues.  In general I've had mixed results doing in-place upgrades whether it's Redhat, Suse or Ubuntu.  Sometimes there are easy fixes to issues that popup other times I've had to do a complete wipe and reinstall.  While it doesn't solve this issue, Bright 9 has a feature called cm-image that makes it easier to configure different Linux distributions or different versions of a distribution on a single cluster.  It builds the default image, creates a new category and the /cm/shared directory without having to standup a separate head node.