01 / BUILD
Reproducible builds
Compile packages, validate releases and run bounded CI jobs without depending on one machine.
Distributed compute for useful work
Small nodes. Shared capacity. Useful jobs.
OpenMesh Compute pools rented server capacity into a lightweight worker network for reproducible builds, batch processing, mirrors and other bounded open-source workloads.
Worker meshonline
worker-eu-a
queue
worker-eu-b
artifact-cache
reserve
Why this exists
Not every useful workload needs a giant cloud.
A few independent VPS nodes are enough for many background jobs — and spreading them across providers or regions makes maintenance and failures easier to absorb.
02 / BATCH
Batch processing
Run indexing, validation, transforms and scheduled CPU workloads across independent workers.
03 / CACHE
Artifact caching
Keep reusable public build outputs close to the worker mesh and reduce repeated upstream transfers.
04 / SPARE
Reserve capacity
Maintain another worker in a separate region so maintenance does not stop the entire queue.
Workloads
Designed for bounded background jobs.
The network is meant for jobs that can be queued, repeated and audited — not permanent application hosting or arbitrary public compute.
Open-source release jobs
Package builds, checksums, test matrices and reproducibility checks for small projects.
Data preparation
Indexing, conversion, validation and other scheduled CPU-bound transforms.
Mirrors and caches
Temporary copies of public artifacts and datasets to reduce repeated origin traffic.
Network measurements
Low-rate availability and latency checks between worker regions to understand path quality.
Network model
A queue, independent workers, shared results.
Keep tasks simple: describe the job, assign it to an available worker, collect artifacts and logs, then reuse results from the cache where possible.
Describe the job
Inputs, resource limits and expected artifacts are declared before execution.
Choose a worker
The queue picks available capacity based on the task profile.
Run in isolation
The worker executes the bounded task and records the result.
Store artifacts
Outputs and logs return to shared storage or the artifact cache.
Support the nodes
Contributions pay for infrastructure, not subscriptions.
The project is funded by small voluntary contributions. The budget goes toward VPS invoices, transfer, NVMe storage, snapshots and spare capacity in another region.
There are no premium accounts or paid compute tiers. Contribution examples below are illustrative rather than fixed service prices.
See the budget model€5example
Traffic & storageHelps cover transfer, snapshots and short-lived build artifacts.
€15example
Small worker budgetContributes toward a lightweight VPS in the monthly worker pool.
€30example
More CPU & memoryHelps fund a stronger worker for longer or more parallel batch jobs.
€60example
Reserve node capacityContributes toward keeping another region available during maintenance or outages.
OpenMesh Compute is presented as an independent community infrastructure project. Contributions are voluntary and are not described as tax-deductible charitable donations.
Principles
Useful, boring infrastructure.
No invented partner logos, no fake customer counters, no vague “AI platform” claims — just a clear reason for servers to exist and a plausible way to fund them.
Is this a commercial cloud?
No. The project does not sell VPS instances or public user accounts. It maintains a shared worker pool for project workloads.
Can anyone run arbitrary code?
No. Workloads are reviewed and bounded before being accepted into the worker queue.
Why multiple servers?
Independent nodes provide parallel capacity, regional diversity and room for maintenance without stopping every job.
Where does support money go?
Into infrastructure expenses: compute, transfer, storage, snapshots and spare worker capacity.