lib.itmens/docs/configuration.md

# Configuration


## Important settings you may want to change first

absolutely set these in `.env` before start the instance for the first time:

 - `NEODB_SECRET_KEY` - 50 characters of random string, no white space
 - `NEODB_SITE_NAME` - the name of your site
 - `NEODB_SITE_DOMAIN` - the domain name of your site

**`NEODB_SECRET_KEY` and `NEODB_SITE_DOMAIN` must not be changed later.**

if you are doing debug or development:

 - `NEODB_DEBUG` - True will turn on debug for both neodb and takahe, turn off relay, and reveal self as debug mode in nodeinfo (so peers won't try to run fedi search on this node)
 - `NEODB_IMAGE` - the docker image to use, `neodb/neodb:edge` for the main branch

## Settings for customization

 - `NEODB_SITE_LOGO`
 - `NEODB_SITE_ICON`
 - `NEODB_USER_ICON`
 - `NEODB_SITE_COLOR` - one of [these color schemes](https://picocss.com/docs/colors)
 - `NEODB_SITE_INTRO`
 - `NEODB_SITE_HEAD`
 - `NEODB_SITE_DESCRIPTION`
 - `NEODB_PREFERRED_LANGUAGES` - preferred languages when importing titles from 3rd party sites like TMDB and Steam, comma-separated list of ISO-639-1 two-letter codes, `en,zh` by default. It can includes languages with no UI translations yet, e.g. if set to `ja,en,zh`, NeoDB scraper will fetch catalog metadata in three languages if they are available from third party sites, Japanese users (= whose browser language set to ja-JP) will see English UI with Japanese metadata.
 - `NEODB_DISCOVER_FILTER_LANGUAGE` - `False` by default; when set to `True`, `/discover/` will only show items with languages match one of `NEODB_PREFERRED_LANGUAGES`.
 - `NEODB_DISCOVER_SHOW_LOCAL_ONLY` - `False` by default; when set to `True`, only show items marked by local users rather than entire network on `/discover/`
 - `NEODB_DISCOVER_UPDATE_INTERVAL` - minutes between each update for popular items on `/discover/`
 - `NEODB_SITE_LINKS` - a list of title and links to show in the footer, comma separated, e.g. `Feedback=https://discord.gg/8KweCuApaK,ToS=/pages/rules/`
 - `NEODB_INVITE_ONLY` - `False` by default, set to `True` to require invite code(generated by `neodb-manage invite --create`) to register
 - `NEODB_ENABLE_LOCAL_ONLY` - `False` by default, set to `True` to allow user to post marks as "local public"
 - `NEODB_LOGIN_MASTODON_WHITELIST` - a list of Mastodon instances to allow login from, comma separated
 - `NEODB_EMAIL_FROM` - the email address to send email from
 - `NEODB_EMAIL_URL` - email sender configuration, e.g.
 	- `smtp://<username>:<password>@<host>:<port>`
 	- `smtp+tls://<username>:<password>@<host>:<port>`
 	- `smtp+ssl://<username>:<password>@<host>:<port>`
 	- `anymail://<anymail_backend_name>?<anymail_args>`, to send email via email service providers, see [anymail doc](https://anymail.dev/)
 - `DISCORD_WEBHOOKS` - Discord channel to send notification about user submitted suggestion and changes, e.g. `suggest=https://discord.com/api/webhooks/123/abc,audit=https://discord.com/api/webhooks/123/def`. Both suggest and audit channels must be in forum mode.


## Settings for Federation

 - `NEODB_SEARCH_PEERS` is empty by default, which means NeoDB will search all known peers running production version of NeoDB when user look for items. This can be set to a comma-separated list of host names, so that NeoDB will only search those servers; or search no other peers if set to just `-`.

 - `NEODB_DISABLE_DEFAULT_RELAY` is set to `False` by default, the server will send and receive public posts from `relay.neodb.net`.

 	`relay.neodb.net` is [open sourced](https://github.com/neodb-social/neodb-relay) and operated by NeoDB developers, it works like most ActivityPub relays except it only relays between NeoDB instances, it helps public information like catalogs and trends flow between NeoDB instances. You may set it to `True` if you don't want to relay public posts with other NeoDB instances.

## Settings for external item sources

- `SPOTIFY_API_KEY` - base64('CLIENT_ID:SECRET'), see [spotify doc](https://developer.spotify.com/documentation/web-api/tutorials/client-credentials-flow)
- `TMDB_API_V3_KEY` - API v3 key from [TMDB](https://developer.themoviedb.org/)
- `GOOGLE_API_KEY` - API key for [Google Books](https://developers.google.com/books/docs/v1/using)
- `DISCOGS_API_KEY` - personal access token from [Discogs](https://www.discogs.com/settings/developers)
- `IGDB_API_CLIENT_ID`, `IGDB_API_CLIENT_SECRET` - IGDB [keys](https://api-docs.igdb.com/)
- `NEODB_SEARCH_SITES` is empty by default, which means NeoDB will search all available sources. This can be set to a comma-separated list of site names (e.g. `goodreads,googlebooks,spotify,tmdb,igdb,bandcamp,apple_podcast`), so that NeoDB will only search those sites; or not search any of them if set to just `-`.


## Other maintenance tasks

Add alias to your shell for easier access

```
alias neodb-manage='docker-compose --profile production run --rm shell neodb-manage'
```

Toggle user's active, staff and super user status

```
neodb-manage user active <username>
neodb-manage user staff <username>
neodb-manage user super <username>
```

Create an invite link

```
neo-manage invite --create
```

Manage user tasks and cron jobs

```
neodb-manage task --list
neodb-manage cron --list
```

Manage search index
```
neo-manage index --reindex
```

Crawl links
```
neodb-manage cat [--save] <url>  # parse / save a supported link
neodb-manage crawl <url>  # crawl all recognizable links from a page
```


## Run without Docker

It's currently possible but quite cumbersome to run without Docker, hence not recommended. However it's possible to only use docker to run neodb server but reuse existing PostgresQL/Redis/Typesense servers with `compose.override.yml`, an example for reference:

```
services:
  redis:
    profiles: ['disabled']
  typesense:
    profiles: ['disabled']
  neodb-db:
    profiles: ['disabled']
  takahe-db:
    profiles: ['disabled']
  migration:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  neodb-web:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
    healthcheck: !reset {}
  neodb-web-api:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
    healthcheck: !reset {}
  neodb-worker:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  neodb-worker-extra:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  takahe-web:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  takahe-stator:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  shell:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  root:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  dev-neodb-web:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  dev-neodb-worker:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  dev-takahe-web:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  dev-takahe-stator:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  dev-shell:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
  dev-root:
    extra_hosts:
      - "host.docker.internal:host-gateway"
    depends_on: !reset []
```
(`extra_hosts` is only needed if PostgresQL/Redis/Typesense is on your host server)


## Multiple instances on one server

It's possible to run multiple clusters in one host server with docker compose, as long as `NEODB_SITE_DOMAIN`, `NEODB_PORT` and `NEODB_DATA` are different.


## Scaling up

For high-traffic instance, spin up these configurations to a higher number, as long as the host server can handle them:

 - `NEODB_WEB_WORKER_NUM`
 - `NEODB_API_WORKER_NUM`
 - `NEODB_RQ_WORKER_NUM`
 - `TAKAHE_WEB_WORKER_NUM`
 - `TAKAHE_STATOR_CONCURRENCY`
 - `TAKAHE_STATOR_CONCURRENCY_PER_MODEL`

Further scaling up with multiple nodes (e.g. via Kubernetes) is beyond the scope of this document, but consider run db/redis/typesense separately, and then duplicate web/worker/stator containers as long as connections and mounts are properly configured; `migration` only runs once when start or upgrade, it should be kept that way.