Products
In preparation
All three are being assembled now. No pre-orders, no placeholder prices — they go live when they are actually usable.
Slurm Production Playbook
slurm.conf templates for CPU+GPU heterogeneous nodes, partition and QoS design, preemption policy, and a failure-code cheat sheet from real incidents.
Get notified →Cluster Delivery Checklist
Rack-to-acceptance checklist: cabling and IB topology, firmware and driver lifecycle, burn-in benchmarks, acceptance criteria and sign-off sheet.
Get notified →Observability Templates
Prometheus + Grafana dashboards for GPU clusters: node health, IB link errors, job queue pressure and the alerts worth waking up for.
Get notified →How these are made
- From the field, not from docs. Each item is distilled from clusters I actually built, tuned or repaired. Client and employer details are stripped; the engineering decisions and numbers stay.
- Copy-paste first. If something cannot be applied without rewriting it, it is not finished.
- Free samples on the site. Large parts of the method are already public in the 15-chapter knowledge base — the paid material is the condensed, ready-to-run version.
Why paid
Writing and validating this takes time that does not come back. Charging for the packaged version is what makes the free chapters keep coming.
Want one — or want me to build something for your setup?
If a product does not cover your case, work with me instead: a review or a scoped engagement is usually faster than adapting a template.