There are two recommended ways to deploy Sourcegraph:

• Single-machine Docker deployment
• Multi-machine Kubernetes deployment

But what if your organization wants a multi-machine deployment without using Kubernetes? What if you use a different container management platform, for example? This project aims to solve that, by providing a pure-Docker deployment option.

The goal is that anyone using a container management platform other than Kubernetes (Netflix's Titus, Apache's Mesos, etc.) would be able to use this repository as a reference for how to deploy Sourcegraph.

First clone the repository, then:

./deploy.sh

Visit http://localhost:3080 to visit the running Sourcegraph instance!

./teardown.sh

To understand the system topology:

1. Look at deploy.sh to get an overview of services.
2. Every service (deploy-*.sh) has documentation inline indicating:
   • What the service does / provides.
   • What ports the service exposes.
   • What other services it talks to (see environment variables).

Every service (deploy-*.sh) documents inline what the system requirements are (CPU/Memory/Disk allocation). For example, the frontend service.

To scale the cluster deployment, you will need to:

1. Deploy more instances of gitserver, indexed-search, searcher and symbols services as desired.
2. Configure the frontend and frontend-internal to communicate with the new instances.
3. You're done! You do not need to configure or restart any other services.

This deployment comes with metrics and tracing built-in. See metrics and tracing for details.

See https://docs.sourcegraph.com/admin/nginx#nginx-for-other-sourcegraph-clusters-e-g-pure-docker

1. Ensure your Site admin > External services code host configuration is configured to use SSH. For example, by setting the gitURLType field to "ssh". You may still need to configure an access token or other codehost authentication method in order for Sourcegraph to discover your repositories.

Alternatively, you may use the OTHER codehost type under External services, which allows you to directly specify Git repository URLs for cloning.

2. Provide your gitserver instances with your SSH / Git configuration (usually just .ssh/id_rsa, .ssh/id_rsa.pub, and .ssh/known_hosts -- but you can also provide other files like .netrc, .gitconfig, etc. if needed) by mounting it into the sourcegraph users home directory in the gitserver containers. For example, by adding the following flag:

-v ~/my-sourcegraph-ssh-config/.ssh:/home/sourcegraph/.ssh

All future Git cloning operations will use the credentials configured there.

If you wish, you can test that cloning with your configuration is working by performing the clone in a gitserver container shell, e.g. first acquire shell access:

$ docker exec -it gitserver-0 sh

Then try cloning the repository:

$ git clone ssh://git@myhost.com/my/repo /tmp/my-repo

Open an issue (in sourcegraph/sourcegraph) or contact us (support@sourcegraph.com), we are happy to help!