Hey everyone!
We're currently in the midst of a large-scale in-place upgrade campaign for our RHEL 7 and Windows Server 2012 (both R2 and standard versions) to decrease our obsolescence numbers. So far, we've been focusing exclusively on virtual machines using an Ansible playbook that manages everything from snapshots to repo configurations. We've just crossed the milestone of upgrading over 1000 servers!
I'm curious to know how common this approach is. How do you all manage your obsolescence? Also, keep in mind that most of our applications are Java-based, which allows us to isolate the OS version using the JVM.
Thanks a ton for sharing your experiences!
4 Answers
I'm kind of new to Ansible myself. Could someone share what their playbooks look like? I'm wondering how they check for successful upgrades and whether they mount ISOs or handle snapshots with Ansible.
If it were me at that scale, I'd consider treating the VMs as disposable. Why not use Terraform to blow everything away and redeploy fresh services instead of doing in-place upgrades?
We actually don't do in-place upgrades. Instead, we rely heavily on Ansible to deploy new servers, then developers transition to the new setup while we decommission the old ones.
Honestly, it sounds like you're a bit behind on the whole migration away from RHEL7. A lot of organizations have been steering clear of that for a couple of years now!
Yeah, same here! My place has had policies against those upgrades for a while. It's all about keeping up with security and support, after all.
Good point! Our pipeline combines Terraform with Ansible, so we can spin up servers pretty quickly. The downside is that our app team isn't quite ready to automate their deployments yet, which slows us down a bit.