Skip to content

Install with Docker Compose

Maki publishes an image for amd64 and arm64 at ghcr.io/orbitmpgh/maki:latest.

services:
maki:
image: ghcr.io/orbitmpgh/maki:latest
container_name: maki
environment:
- PUID=1000
- PGID=1000
volumes:
- ./maki-config:/config
- /path/to/manga-library:/library
ports:
- "8990:8990"
mem_reservation: 512m
restart: unless-stopped

PUID and PGID set the user the container runs as. Maki fixes the ownership of /config on every start but never touches /library, so that user needs write access to your library folder already. See Volumes, PUID and PGID.

  1. Run docker compose up -d, open http://localhost:8990 and create the administrator account.
  2. The setup guide opens next. Add /library as a library folder there. Maki downloads the local MangaBaka database in the background (about 350 MB, 3.5 GB on disk) and finishes opening once it is installed.
  3. Continue with Add your first series.

mem_reservation is a soft limit: when the host runs short of memory, the kernel takes Maki’s file cache back first. Most dashboards count that cache as the container’s memory, so without a reservation Maki can look much bigger than it is.

Don’t set a hard mem_limit below 1.5 GB yet, since a scrape and a search running together can briefly need close to 1 GB.

The compose file for Kavita, FlareSolverr and Suwayomi, and where each is connected in Maki, is on Kavita, FlareSolverr and Suwayomi.