back to field notes blog.ahcomputing.com
2025-04-26 ~6 min read

Proxmox Post-Mortem: What I'd Do Differently

Two nodes, one cluster, several expensive lessons. A frank accounting of what broke, what held, and what I'm rebuilding from scratch with better intentions.

Background

Write your intro here. Set the scene — what did the cluster look like when you built it, what were your goals at the time, and what did you not know yet.

grafzeppelin (10.1.111.100) is an E3-1231 v3 with 32GB of RAM — the main workhorse. peterstrasser (10.1.111.110) is a dual-socket E5620 in the office. They cluster fine. Clustering fine doesn't mean you thought it through.

What Broke

Write about the failures here. Be specific — readers remember specifics. What did you lose? What recovered? What didn't?

Replace this with the thing you wish someone had warned you about.

What Held

Give credit where it's due. What about your setup actually worked when things went sideways?

What I'd Do Differently

The actual value in a post-mortem. Three to five concrete things — not abstract advice, real changes.

# Example: what your backup config looks like now vs. then
# Before: nothing
# After:
vzdump 100 --storage backups --compress zstd --mode snapshot

Closing thought. What does this cluster look like now, and what's the next thing you're watching for?