The Pi Swarm Migration: A Progress Check, Not a Victory Lap
A measured look at the home-lab move: restored applications, repaired public routes, monitoring that follows the services, and the work still left to do.
-
-
The Swarm Migration Is Working—But It Isn’t Finished Yet
An honest field report from moving a self-hosted Docker estate onto a four-node Raspberry Pi Swarm: what moved, why Paperclip took two hours, what is currently degraded, and what remains.
-
Sonarr recovery without handholding: two unattended fault windows
A failed first attempt, two genuinely hands-off Sonarr recoveries, and the evidence separating a working recovery controller from another manually launched restore script.
-
A Container Is Not a Recovery Plan: Three Stateful Pi Swarm Trials
A four-Pi lab recovered Linkding, PostgreSQL-only and Redis-only test workloads in 64.5, 28.9 and 24.4 seconds—with exact snapshots, fenced writers, verified data, and clear limits.
-
microSD vs. NVMe on a Raspberry Pi 5: We Measured the Difference
Same Pi, two storage devices, three repeatable read tests: our NVMe SSD delivered 4.9× the sequential throughput and 14.9× the concurrent random-read throughput of microSD.
-
Can 74 containers float? A Docker Swarm readiness audit
A container-by-container inspection of both Docker hosts in the estate: which of the 74 containers can be scheduled on any node, and what actually pins the other 64 to one host.
-
Draft Recommendation: Rebuild Oconto’s Docker Platform Around a Clean Local Swarm
Updated recommendation for two Raspberry Pi 5s and two Raspberry Pi 4s: three active Swarm managers may also carry carefully limited lightweight ARM workloads, while heavy and stateful work stays off the control plane.
