Skip to content

Latest commit

 

History

History
185 lines (134 loc) · 8.93 KB

container_usage.md

File metadata and controls

185 lines (134 loc) · 8.93 KB
Raw

🐳 Running Jikan API in a container

The most easiest way to get started is to use our container setup cli script after checking out the repo with git (linux only):

./container-setup.sh start

This will:

  • Prompt you for the required passwords and usernames
  • Sets up a production ready setup with redis, typesense and mongodb (almost same as the public api at api.jikan.moe)
  • Sets mongodb to use max 1gb of memory
  • Configures jikan-api to add CORS headers to responses.

Note: The script supports both docker and podman. In case of podman please bare in mind that sometimes the container name resolution doesn't work on the container network. In those cases you might have to install aardvark-dns package. On Arch Linux podman uses netavark network by default (in 2023) so you will need to install the before mentioned package.

Note 2: The script will start the jikan API, but if you start it for the first time, it won't have any data in it! You will have to run the indexers through artisan to have data. See "Running the indexer with the script" section.

The script has the following prerequisites and will notify you if these are not present:

  • git
  • docker or podman
  • docker-compose or podman-compose

Available commands in the cli script

============================================================
Jikan API Container Setup CLI
============================================================
Syntax: ./container-setup.sh [command]
---commands---
help                   Print CLI help
build-image            Build Image Locally
start                  Start Jikan API (mongodb, typesense, redis, jikan-api workers)
stop                   Stop Jikan API
validate-prereqs       Validate pre-reqs installed (docker, docker-compose)
execute-indexers       Execute the indexers, which will scrape and index data from MAL. (Notice: This can take days)
index-incrementally    Executes the incremental indexers for each media type. (anime, manga)

Running the indexer with the script

When you first startup the app you will have an empty database. To fill it up you can execute the following command:

./container-setup.sh execute-indexers

Please note that this command can take 4-5 days to run. You can run it in the background with the & marker:

./container-setup.sh execute-indexers &

If interrupted then you will have to manually resume the indexing, otherwise the above command will just start again from the beginning.

Updating to a newer version

You need to stop the app first:

./container-setup.sh stop

Then remove the jikan-api image from your local storage and pull the new one. Set the JIKAN_API_VERSION environment variable to the latest image tag. This can be either latest or the version v4.0.0-11.

JIKAN_API_VERSION=latest ./container-setup.sh start

More customised setups

Some of you might only want to run the jikan-rest app with only mongodb, without the more sophisticated search functionality. In those cases we don't have a docker-compose config for you. You need to start the jikan-rest container with atleast a mongodb instance. The jikan-rest container will require a .env file mounted where you configure the credentials for mongodb.

docker run -d --name=jikan-rest -p 8080:8080 -v ./.env:/app/.env jikanme/jikan-rest:latest
  • Container listens on port 8080 for http requests
  • By mounting your .env file on the container via -v ./.env:/app/.env command line option for docker run you can configure Jikan API.

Important: You need to either mount a .env file on the container or specify the configuration through environment variables to make Jikan API work in the container. Jikan API needs a MongoDB and optionally a search engine. In high load environments additionally a redis server is required too. The configuration should point to the correct address of these services.

Tip: If you run the container on a non-default network, you can use the container names in the configuration to specify the address of services like MongoDB and TypeSense.

There is also a Dockerfile in the repo which you can use to build the container image and startup the app in a container:

docker build -t jikan-rest:nightly .
docker run -d --name=jikan-rest -p 8080:8080 -v ./.env:/app/.env jikan-rest:nightly

Most of the time it's enough to just use the image from Docker Hub.

Docker compose usage

docker-compose up

This does the same thing as the container-setup.sh script mostly, but you will have to create the secret files yourself. The following secret files are required for credentials (put them next to the docker-compose.yml file):

  • db_admin_password.txt
  • db_admin_username.txt
  • db_password.txt
  • db_username.txt
  • redis_password.txt
  • typesense_api_key.txt

You can customise the Jikan API config through ./docker/config/.env.compose file. (E.g. you don't want CORS headers)

Please note: The syntax rules of docker compose for .env applies here: https://docs.docker.com/compose/env-file/#syntax-rules

Additional configuration: You can change the mongodb memory usage via MONGO_CACHE_SIZE_GB environment variable. It sets how many gigabytes of memory is available for wired tiger. Default is 1. This is useful for systems with low memory capacity.

Note for Podman

If you build the container image yourself with podman, the resulting image format will be OCI by default. To make the health checks work in that situation you need to run the container the following way:

podman run -d --name=jikan-rest -p 8080:8080 -v ./.env:/app/.env --health-start-period=5s --health-cmd="curl --fail http://localhost:2114/health?plugin=http || exit 1" jikan-rest:nightly

Configuration of the container

You can also change the settings of Jikan through setting environment variables via the -e command line argument option for the docker run command. These environment variables are the same as the options found in the .env file. We also provide a sample file called .env.dist.
Additionally, you can use the --env-file option of docker run to specify configuration for Jikan, in which case you put all the configuration in the env file.

docker run -d --name=jikan-rest -p 8080:8080 --env-file ./env.list jikanme/jikan-rest:latest

The env-file should contain env var value pairs line by line.

VAR1=value1
VAR2=value2

There are additional configuration options:

Name Description
RR_MAX_WORKER_MEMORY (Number) Configures the available memory in megabytes for the php scripts
RR_MAX_REQUEST_SIZE_MB (Number) Configures the max allowed request body size in megabytes
JIKAN_QUEUE_WORKER_PROCESS_NUM (Number) Configures the number of running queue worker processes. (You want to increase this if you experience huge load)
JIKAN_ENABLE_PERIODICAL_FULL_INDEXER (Bool) Configures whether to run the anime/manga indexer every week, which would crawl all anime/manga at first then it would just grab the latest anime/manga entries from MAL. Defaults to false.

You can read more about additional configuration options on the Configuration Wiki page.

Some facts about the container image

  • Jikan uses RoadRunner as an application server within the container.
  • Both wget and curl exists in the container image.
  • The script in docker-entrypoint.php sets safe defaults. Because of this by default the app won't behave the same way as the publicly available version of the app at https://api.jikan.moe/v4. The default settings:
    • No redis caching
    • No search index usage (inaccurate search results)
  • Via Roadrunner multiple processes are running in the container, and their logs are aggregated and forwarded to stdout.
    • These processes are:
      • the php processes ingesting the http requests
      • Supercronic, which runs cron jobs.
      • Queue workers for populating the search index and other background jobs.