Search for open source employee scheduling software and you’ll find three kinds of result: abandoned repositories, general-purpose calendar apps being stretched into a roster, and commercial products with “free plan” pages ranking for the term. Actual maintained, self-hostable shift scheduling software is a much shorter list than the search results suggest.
This is that shorter list, plus the part most roundups skip: what self-hosting one actually costs once you count the hours.
Disclosure: this is the Manage Roster blog and Manage Roster is closed-source hosted software. It appears once at the end, with the trade-off stated plainly. If self-hosting is a hard requirement for you, the projects above it are your options and we are not one of them.
| If you want… | Consider | Reality check |
|---|---|---|
| A scheduling module inside a full ERP | ERPNext / Frappe HR | Heavy, but genuinely maintained |
| Shift planning inside an HR suite | OrangeHRM (Community) | Roster features are thin |
| Constraint-solved rosters from a library | Timefold (ex-OptaPlanner) | A solver, not an app — you build the app |
| Appointment scheduling, not shift rostering | Cal.com, Easy!Appointments | Different problem — check before adopting |
| To stop maintaining a server | A hosted tool with a free tier | Licence cost isn’t the whole cost |
A large share of “open source scheduling” results solve appointment booking — a client picks a slot from your availability. Cal.com and Easy!Appointments are good at that.
Shift rostering is close to the opposite problem: you assign staff to fixed coverage windows, subject to leave, skills, rest rules and cost. Adopting an appointment tool for a roster is one of the most common wasted evaluations in this category, because the demo looks adjacent and the model doesn’t fit.
If your question is “who is covering 22:00–06:00 on Thursday”, you need the second kind. Nothing in the first category will answer it.
The most substantial genuinely open source option, because scheduling sits inside a maintained ERP with real development behind it. You get shift types, shift assignment, attendance, leave and payroll in one system, and it’s actively developed with a commercial company sustaining it — which is the single best predictor that a project will still be alive in three years.
The cost: it’s an ERP. You’re adopting a large system to get a roster, and configuration is a project. Sensible if you want the rest of the suite; disproportionate if you only want to know who’s on nights.
A long-established open source HR system. Solid on leave, employee records and time tracking. Its shift and roster capability is comparatively thin, and several of the more advanced workforce features sit in the paid edition rather than the community one — worth checking the current feature split against your list before committing.
The most interesting entry and the most misunderstood. It is a constraint solver, not a scheduling application: you express your rules — coverage requirements, rest periods, skill matching, fairness — and it optimises assignments against them. Employee rostering is one of its canonical example problems.
If your rostering problem is genuinely hard — hundreds of staff, complex skills, competing fairness constraints — nothing on a free tier of a SaaS product will match what a solver can do.
The cost: you are building the application. Timefold gives you an engine; the UI, data model, auth, notifications and mobile experience are yours. Budget in developer-months.
Several projects that rank well for this term are effectively unmaintained — Staffjoy is the best-known, open sourced after the company wound down. There’s real value in reading their code, and real risk in running it: no security patches, no dependency updates, no upstream.
Check three things before adopting any of them: date of the last commit, whether open issues get responses, and whether more than one person has commit rights. A project with one contributor and a quiet six months is a liability you’re volunteering for.
“Free” refers to the licence. Here’s the rest of it, for a small team:
| Line | Realistic figure |
|---|---|
| VPS (2GB, enough for a small deployment) | $10–20/month |
| Domain + TLS | ~$15/year (Let’s Encrypt handles TLS) |
| Backups (offsite, tested) | $5/month |
| Initial setup and configuration | 8–20 hours |
| Ongoing patching, updates, breakage | 2–4 hours/month |
| Year one, hours valued at $50 | ~$1,200–2,600 |
Compare that against a hosted tool at $2–9 per user per month and the maths only favours self-hosting at larger headcounts — or when data residency is a hard requirement rather than a preference, which is a legitimate reason that has nothing to do with cost.
The line people underestimate is the last recurring one. It isn’t the setup; it’s the Tuesday eighteen months later when a dependency update breaks login and you’re the only person who knows how the thing was deployed.
The failure mode for self-hosted internal tools is almost never technical. It’s that the person who set it up leaves.
Before adopting, answer honestly: who is the second person who can restore this from backup? If the answer is nobody, you don’t have a self-hosted tool, you have a dependency on one colleague. That’s a worse operational position than a SaaS subscription, not a better one.
If you land on “no capacity to self-host”, the honest alternative is a hosted tool with a real free tier. We covered those in the best free staff roster apps — Sling is free to 30 users, Homebase free for one location, Findmyshift free to 5.
Manage Roster is our own, and closed-source hosted — the opposite trade to this article’s subject. You give up data residency and the ability to modify it; you get a 24-hour day view, live overtime events, per-person holiday calendars, coverage-gap detection and an MCP server, without owning a server. Free during beta for up to 5 agents with code BETA2026, $29/month on Pro for unlimited agents.
If self-hosting is a requirement rather than a cost-avoidance strategy, that trade won’t appeal, and it shouldn’t. ERPNext is where we’d point you.
Is there a good open source alternative to When I Work or Deputy? Not a direct one. Those products’ value is heavily in polish, mobile apps and integrations — the parts open source projects are weakest on. ERPNext is the closest functional match, but it is an ERP rather than a focused scheduling app, and the experience reflects that.
Can I self-host something free and still get a mobile app? Rarely. Most of these projects are responsive web at best. For shift workers this matters more than feature lists suggest — a roster that’s awkward on a phone gets photographed and shared in a group chat, and you’re back to a shadow schedule.
What about Odoo? Odoo has planning and attendance modules and a community edition. Same caveat as ERPNext — you’re adopting a business suite to get a roster — plus a sharper split between community and enterprise features. Check which edition contains the parts you need before planning around it.
Is Timefold overkill for 20 staff? Yes. Constraint solvers earn their complexity when a human can no longer hold the problem in their head. At 20 people with conventional shifts, a spreadsheet beats a solver on total cost, and a $29/month tool beats both on time.
How do I evaluate whether a project is maintained? Last commit date, response time on open issues, and number of people with commit rights. One contributor and six quiet months means you are adopting the maintenance burden, whatever the README says.
Does open source mean more secure? It means auditable, which is not the same thing. A self-hosted instance you don’t patch is less secure than a managed service someone patches for you. Security follows from who does the work.
The honest summary of open source shift scheduling in 2026: ERPNext if you want a maintained system and can absorb an ERP, Timefold if your problem is genuinely a constraint-optimisation problem and you have engineers, and careful scrutiny of anything else’s commit history before you trust a roster to it.
Everything else being equal, self-hosting is the right call when data residency is a requirement or you have real platform capacity. It’s the wrong call when the goal is avoiding a $30 monthly invoice — that’s a saving you repay in hours, at a worse rate, forever.
For the hosted side of the comparison, see the 6 best roster management software tools and staff scheduling software vs spreadsheets.
Project capabilities and community/enterprise feature splits change frequently in this category. Verify against each project’s current documentation before committing — figures and feature boundaries here reflect September 2026.