Job builder-web on ubuntu-20.04


Skip to builds

Builder-web - a web frontend for reproducible builds

Builder-web takes in submissions of builds, typically from builder, and displays the produced artifacts in a way that makes it easy to compare checksums and build status. Produced binaries can be downloaded and executed. itself runs builder-web.


Builder-web is a single binary web server using a sqlite3 database with versioned schemas. Finished builds from builder are uploaded to builder-web, stored and indexed in the database and presented in the web interface to the user. Users can:

  • Get an overview of jobs - a job is typically script or opam package that is run and builds an artifact,
  • Browse all builds of a job - each job may have multiple builds, that are executed periodically by builder
  • Browse individual build and download artifacts and build information for reproducing the same binary.
  • Compare two builds, observing the differences in used opam packages, environment variables, and system packages.
  • Search for the SHA-256 hash of a binary to view a build that resulted in that binary.


Installing from source can be done with opam: opam install builder-web.

We also provide reproducible binary packages.


Builder-web consists of a binary builder-web that runs a web server on port 3000 listening on all interfaces by default. These values can be changed with the --port and --host flags respectively. See builder-web --help for more information.

Service scripts for FreeBSD and systemd are provided.

The web server expects a sqlite3 database in its data directory. An empty database can be created with builder-db migrate.

Database migrations

The sqlite3 database builder-web uses contains versioning information. On every schema change the database schema version is updated, and migration and rollback scripts are provided. The tool for database migrations is builder-migrations. See the builder-migrations --help output for each migration for further details.

Less common workflows

Here are listed some less common but useful workflows:

Extracting builds from one server to another

This is useful for development on a separate machine that doesn't run the build jobs itself.

On the source server:

builder-db extract-build <build-hash> --dest <build-hash>.full

After copying the file over the destination server (you need a user first, see builder-db user-add --help):

curl --data-binary @<build-hash>.full http://<user>:<passwd>@localhost:<builder-web-port>/upload


Back to readme

Including failed builds here.