← Complete project index

Infrastructure & Operations / Built / working

Distributed Compute & Build Fabric

Nomad-based compute path across multiple machines with deterministic build execution and fallback capacity rather than tying jobs to one workstation.

STATUSBuilt / working
AREAInfrastructure & Operations
STACK4 documented technologies
INTEGRATIONS3 connected / bounded services
01

OVERVIEW

What I built.

Nomad-based compute path across multiple machines with deterministic build execution and fallback capacity rather than tying jobs to one workstation.

01

Built a Nomad-based execution path across the laptop, desktop and home server environment.

02

Used Tailscale connectivity and worker selection so jobs are not permanently tied to one machine.

03

Applied it to repeatable build/deployment work where fallback compute is useful.

02

SYSTEM ARCHITECTURE

How the system is divided.

01
Architecture boundary

Controller/scheduler -> Nomad clients on available machines -> isolated job execution.

02
Architecture boundary

Tailscale provides stable private reachability between participating hosts.

03
Architecture boundary

Work can move between eligible nodes while state/intent remains in the control layer.

03

NETWORK / DATA FLOW

What talks to what.

04

TECHNOLOGY STACK

The technical surface.

01Nomad
02Linux
03Tailscale
04Build automation
05

IMPLEMENTATION

How it works in practice.

  1. 01

    Controller/scheduler -> Nomad clients on available machines -> isolated job execution.

  2. 02

    Tailscale provides stable private reachability between participating hosts.

  3. 03

    Work can move between eligible nodes while state/intent remains in the control layer.

06

TESTING & VALIDATION

How I know it works.

TEST 01

Nomad client registration on multiple machines

TEST 02

Controller/compute smoke tests

TEST 03

Cross-node reachability over Tailscale

TEST 04

Fallback node behavior

TEST 05

Build/deployment task execution

07

EXTERNAL SERVICES & DEPENDENCIES

What sits outside the core.

COREDistributed Compute & Build Fabric
08

ENGINEERING EVIDENCE

Artifacts behind the claims.

01Nomad client installs
02Smoke-test runs
03Multi-host topology
04Fallback execution policy
05Build-worktree usage
09

BOUNDARIES / CONSTRAINTS

What the project does not pretend to solve.

Capacity depends on which machines are online

State should live in the control layer rather than a transient worker

Not every workload is suitable for every node

10

RESULT / CURRENT STATE

What this produced.

A small distributed execution fabric that reduces dependence on a single always-on workstation.