1
0
Fork 0
mirror of https://gitlab.com/hagrid-keyserver/hagrid.git synced 2023-02-13 20:55:02 -05:00
No description
Find a file
2019-03-06 14:10:44 +01:00
contrib Use template systemd service file 2019-02-03 12:43:13 +00:00
database Move the database module to a new crate. 2019-03-05 17:30:56 +01:00
dist Link to Hagrid's source, and the revision. 2019-03-06 14:10:44 +01:00
src Limit size of uploads. 2019-03-06 13:40:36 +01:00
.gitignore ignore vim swap files in .gitignore 2019-02-22 16:15:17 +01:00
.gitlab-ci.yml ci: Be quiet, don't update, run tests, don't do release builds. 2019-03-05 17:33:33 +01:00
build.rs fix warning in build.rs 2019-02-22 16:34:45 +01:00
Cargo.toml Move the database module to a new crate. 2019-03-05 17:30:56 +01:00
hagrid-routes.conf Limit size of uploads. 2019-03-06 13:40:36 +01:00
nginx-site.conf split up nginx files 2019-02-27 11:39:38 +01:00
nginx.conf split up nginx files 2019-02-27 11:39:38 +01:00
README.md Implement upload via HKP. 2019-03-05 16:16:02 +01:00
rustfmt.toml use rustfmt to format source 2019-02-08 12:46:31 +01:00

Hagrid

Hagrid is a verifying OpenPGP key server. When a new key is uploaded a token is sent to each user ID via email. This token can be used to verify the user ID. Keys can be queried by their verified user IDs (exact match) and their primary keys fingerprint. Keys can be deleted by clicking a link send to all user IDs.

Quick Start

Building Hagrid required a working Rust nightly toolchain. The key server uses the filesystem to store keys, user IDs and tokens. To run it, supply the absolute path to where you want the database to live and the absolute path to the template directory.

cargo run --bin hagrid -- `pwd`/dist

This will spawn a web server listening on port 8080.

Hagrid uses sendmail for mailing, so you also need a working local mailer setup. The FROM field of the mails can be configured with the -F switch.

Usage

HKP

Hagrid implements basic HKP (op=get and op=index) so tools like GnuPG and OpenKeychain can use it directly. The differences to SKS are

  • no support for op=vindex,
  • only exact matches for user IDs are returned (i.e. exact=on is always assumed),
  • op=index returns either one or no keys,
  • all packets that aren't public keys, user IDs or signatures are filtered out.

VKS

Hagrid has it's own URL scheme to fetch keys, verify user IDs and delete keys. It's meant to be machine readable, but it's not a REST API. The following URLs are handled.

  • GET /vks/by-fingerprint/<FINGERPRINT> retrieves the key with the given fingerprint. Hexadecimal digits must be uppercase.
  • GET /vks/by-keyid/<KEY-ID> retrieves the key with the given long key ID. Hexadecimal digits must be uppercase.
  • GET /vks/by-email/<URL-encoded user ID> retrieves the key with the given user ID. Only exact matches are accepted.
  • GET /vks/verify/<token> verifies a user ID using a token string send by email.
  • GET /vks/delete/<fingerprint> requests deletion of the key with the given fingerprint.
  • GET /vks/confirm/<token> confirms a keys deletion request using a token string send by email.

Keys can also be fetched by their subkeys fingerprint and key ID. Note: keys will show up even if no user IDs are verified.

Building

Building Hagrid requires a working nightly Rust toolchain. The easiest way to get the toolchain is to download rustup. After rustup is installed, get the nightly compiler and tools:

cd hagrid
rustup override set nightly

The web server can now be built with the cargo command:

cargo build --release

After compilation a binary is placed in target/release/ called hagrid. The binary is linked statically and can be copied everywhere.

cp target/release/hagrid /usr/local/bin

To deploy the key server copy all directories under public/ to a writable location. Then start the server with the absolute path to the directory as argument:

mkdir /var/lib/hagrid
cp -R dist/* /var/lib/hagrid
hagrid /var/lib/hagrid

This will spawn the server in foreground, listening on 0.0.0.0:8080. The --listen argument can be used to change port and listen address. The server will put all keys and runtime data under the base folder (/var/lib/hagrid in the above example).

Reverse Proxy

Hagrid is designed to defer lookups to reverse proxy server like Nginx and Apache. The key database is a set of 3 directories with static files in them. The directory structure reflects Hagrids URL scheme. This way, lookups via /vks/by-finingerprint, /vks/by-keyid, and /vks/by-email can be handled by (multiple) simple HTTP server(s). A sample configuration for Nginx is part of the repository (nginx.conf).

Community

We're in ##hagrid on Freenode.