Self-hosting Kamra

Own your PMS end to end. Ways to install:

Method	Effort	Best for
Hostinger VPS + one command	Buy a VPS, paste one line	Cheapest path, India / SEA
Docker quickstart on any VPS	One command, three prompts	DigitalOcean, Linode, AWS, your own box
Bench install	Classic Frappe bench	Teams already running Frappe / ERPNext
Frappe Cloud Marketplace	Point and click	Managed Frappe hosting, no server

Every feature is included - there is no paid edition to upgrade to.

Want company books or HR on the same Frappe site? That is optional - see ERPNext and Frappe HR (Marketplace and own-server steps).

Server requirements

	Minimum	Recommended (20+ rooms, POS)
CPU	2 vCPU	4 vCPU
RAM	4 GB	8 GB
Disk	40 GB SSD	80 GB SSD
OS	Ubuntu 22.04/24.04 LTS (any Docker-capable Linux)	-

Measured on our own servers: a full single-property stack (database, cache, workers, web) idles around 600 MB of RAM - 4 GB leaves comfortable headroom for busy days. By default the Docker installer builds the image on your server (20-45 minutes the first time; on boxes under 8 GB it adds a swap file for the build). With access to the prebuilt image it pulls instead - a few minutes, no build memory needed.

Also needed: a domain or subdomain (pms.yourhotel.com) pointed at the server; ports 80/443 open; SSL via Let's Encrypt.

Docker images

Each release is published to ghcr.io/kamra-pms/kamra (linux/amd64). The installer uses it with KAMRA_PULL=1; the package is not public yet, so the server needs docker login ghcr.io with a read:packages token until it is.

Tag	What it is	Use for
stable	The newest release (moves on every release)	What the installer follows; it pins the vX.Y.Z it resolves to
vX.Y.Z	One release, never changes	Pin this in production
nightly	Latest green develop	Testing only

latest is an alias of stable kept for older scripts. The three newest releases plus stable and nightly are kept; older tags are pruned weekly, so pin a recent release and update regularly.

What's in the image

Only public, open-source code: Frappe, payments and Kamra (built from their public GitHub repositories at the release tag, on GitHub Actions - see .github/workflows/release.yml). The image holds no site data and no secrets: your database, uploaded files, site config, passwords and API keys live in Docker volumes and kamra.env on your server, never in the image. Anyone can rebuild the same image from source with KAMRA_BUILD=1.

Typical monthly cost

Provider	Plan that fits	Price	Guide
Hostinger	KVM 2 (2 vCPU / 8 GB)	~₹549/mo	Guide
DigitalOcean	Basic 4 GB	~$24/mo	Guide
Linode	Shared 4 GB	~$24/mo	Guide
AWS	t3.medium + EBS	~$30/mo	Guide

Hyperscaler storefronts (free self-host AMI, planned): AWS / Azure / GCP.

Rather not run a server?

Install from the Frappe Cloud Marketplace - Frappe hosts, backs up and updates it - and add our implementation if you want migration and training done for you. You can export everything and move to self-hosting any time; that's the point of open source.

After install - production checklist

1. Create your property - sign in at /kamra as Administrator with the admin password you chose at install - there is no default - then open /kamra/setup (rooms, room types, rates).
2. Staff users & roles - Hotel Admin / Front Desk / Revenue / Finance / Housekeeping; roles decide what each person sees.
3. Email - set up SMTP for confirmations, invoices and briefings.
4. Payments - add gateway keys in the payments app (e.g. Razorpay Settings); turn off test mode.
5. Scheduler - install.sh enables it; night audit runs at 03:00 site time and housekeeping SLAs escalate every 15 min.
6. Backups - bench backup --with-files on cron to off-site storage.
7. Security - developer_mode 0, strong admin password, HTTPS only.
8. AI (optional) - bring your own model or Connect Claude over MCP; then optionally Connect HeyKoala WhatsApp from Marketplace.

Updating (Docker install)

sudo /opt/kamra/install.sh update

It first takes a full backup (bench backup --with-files) and aborts if the backup fails. Build installs then rebuild the image from the branch recorded in /opt/kamra/apps.json (pin a release with KAMRA_BRANCH=v2.6.4; 10-30 minutes); pull installs pull the newest :stable and pin it (or KAMRA_TAG=vX.Y.Z; a few minutes). Both recreate the containers and run bench migrate on every site. Details: quickstart → Updating.

Bench self-host:

cd frappe-bench/apps/kamra && git pull
bench --site pms.yourhotel.com migrate
bench build && bench restart

Run the eval harness after any update - deterministic checks over money, tax and availability logic:

bench --site pms.yourhotel.com console
>>> from kamra.scripts.eval_harness import execute; execute()