homelab-apps/archivebox.yaml

82 lines
2.9 KiB
YAML

services:
archivebox:
container_name: archivebox
image: archivebox/archivebox:dev
# command: server --quick-init 0.0.0.0:8000
# TODO: hack to workaround https://github.com/ArchiveBox/ArchiveBox/issues/1002
entrypoint: /bin/bash
command: -c "chown -R archivebox:archivebox /app/archivebox/core/migrations && /app/bin/docker_entrypoint.sh server --quick-init 0.0.0.0:8000"
environment:
- ALLOWED_HOSTS=*
- MEDIA_MAX_SIZE=750m
- RESOLUTION=1024,768
volumes:
- /media/mat/passport-5tb/archivebox:/data
labels:
- "traefik.enable=true"
- "traefik.http.routers.archive.rule=Host(`archive.mat`)"
- "traefik.http.routers.archive.entrypoints=web"
- "traefik.http.services.archive.loadbalancer.server.port=8000"
- "traefik.http.routers.archive.service=archive"
- "traefik.docker.network=traefik"
networks:
- traefik
# To run the Sonic full-text search backend, first download the config file to sonic.cfg
# curl -O https://raw.githubusercontent.com/ArchiveBox/ArchiveBox/master/etc/sonic.cfg
# after starting, backfill any existing Snapshots into the index: docker-compose run archivebox update --index-only
# sonic:
# image: valeriansaliou/sonic:v1.3.0
# expose:
# - 1491
# environment:
# - SEARCH_BACKEND_PASSWORD=SecretPassword
# volumes:
# - ./sonic.cfg:/etc/sonic.cfg:ro
# - ./data/sonic:/var/lib/sonic/store
### Optional Addons: tweak these examples as needed for your specific use case
# Example: Run scheduled imports in a docker instead of using cron on the
# host machine, add tasks and see more info with archivebox schedule --help
# scheduler:
# image: archivebox/archivebox:latest
# command: schedule --foreground --every=day --depth=1 'https://getpocket.com/users/USERNAME/feed/all'
# environment:
# - USE_COLOR=True
# - SHOW_PROGRESS=False
# volumes:
# - ./data:/data
# Example: run all your ArchiveBox traffic through a WireGuard VPN tunnel
# wireguard:
# image: linuxserver/wireguard
# network_mode: 'service:archivebox'
# cap_add:
# - NET_ADMIN
# - SYS_MODULE
# sysctls:
# - net.ipv4.conf.all.rp_filter=2
# - net.ipv4.conf.all.src_valid_mark=1
# volumes:
# - /lib/modules:/lib/modules
# - ./wireguard.conf:/config/wg0.conf:ro
# Example: Run PYWB in parallel and auto-import WARCs from ArchiveBox
# pywb:
# image: webrecorder/pywb:latest
# entrypoint: /bin/sh 'wb-manager add default /archivebox/archive/*/warc/*.warc.gz; wayback --proxy;'
# environment:
# - INIT_COLLECTION=archivebox
# ports:
# - 8080:8080
# volumes:
# ./data:/archivebox
# ./data/wayback:/webarchive
networks:
traefik:
external: true