Skip to main content

Docker and docker-compose

Installing A-Parser on Linux and MacOS via Docker is the recommended method; we have prepared an image that includes all necessary dependencies and libraries for A-Parser and Headless Chrome to function.

Before installation​

Downloading the distribution​

Download and unpack the A-Parser distribution:

curl -O https://a-parser.com/members/onetime/ce42f308eaa577b5/aparser-linux-x64.tar.gz
tar zxf aparser-linux-x64.tar.gz
rm -f aparser-linux-x64.tar.gz
note

The download is performed via a one-time link for Linux. You should get it in the Members Area, tab A-Parser -> Downloads

Running via Docker​

Running in foreground mode​

Run A-Parser in foreground mode to see the logs in the terminal:

docker run --rm --name aparser -v $(pwd)/aparser:/app -p 9091:9091 -t aparser/runtime ./aparser -foreground

The -foreground option outputs the log file content directly to the terminal:

Apr 27 07:06:53.10427 [master] foreground mode
Apr 27 07:06:54.06409 [master] Found local Chromium: /app/dist/nodejs/node_modules/puppeteer/.local-chromium/linux-856583/chrome-linux/chrome
Apr 27 07:06:55.36487 [master] NodeJS console: [Rank::CMS] total loaded 1446 apps
Apr 27 07:07:19.01377 [master] Start ProxyChecker nocheck
Apr 27 07:07:19.24418 [master] Start ProxyChecker default
Apr 27 07:07:19.62635 [master] A-Parser v1.2.1176-linux-x64 started, tasks: 0 / 2, memory total: 1989 MB, available: 610 MB

After starting, A-Parser will be available at http://127.0.0.1:9091

Running in the background​

To run the container in the background, use the command:

docker run --rm --name aparser -v $(pwd)/aparser:/app -p 9091:9091 -t -d aparser/runtime ./aparser

Thread tuning when running via Docker​

By default, there is a limit on the number of open files/sockets inside the container. To increase the limit to 10240 when running via Docker, use the --ulimit option:

docker run --rm --name aparser -v $(pwd)/aparser:/app -p 9091:9091 --ulimit nofile=10240:10240 -t -d aparser/runtime ./aparser

Additionally, it is recommended to perform Linux tuning for more threads on the host system.

Updating A-Parser in a Docker container​

Stop the container:

docker stop aparser

Update A-Parser:

docker run --rm --name aparser -v $(pwd)/aparser:/app -p 9091:9091 -t aparser/runtime ./aparser -foreground -doupdate

After the update is complete, start A-Parser

note

You can also use the update via the interface or a one-time link

Running in Headful mode​

To scrape websites that require running the browser in headful mode (with a virtual display), use the aparser/runtime:headful image.

To switch from a regular image to headful:

Stop the container:

docker stop aparser

Download the headful image:

docker pull aparser/runtime:headful

Start A-Parser with the new image:

docker run --rm --name aparser -v $(pwd)/aparser:/app -p 9091:9091 -t -d aparser/runtime:headful ./aparser

For docker-compose, replace the image in docker-compose.yml:

image: aparser/runtime:headful

Then run:

docker compose up -d
note

After starting, enable Chrome headless: 0 in the scraper settings. The log should display Headful when Chrome starts. The novnc server is available on port 5900 for visual monitoring; the password is set via the VNC_PASSWORD variable (default: docker).

Running via docker-compose​

Example docker-compose.yml​

Create a docker-compose.yml file:

version: '3'

services:
  a-parser:
    image: aparser/runtime:latest
    command: ./aparser
    restart: always
    volumes:
      - ./aparser:/app
    ports:
      - 9091:9091

Running via docker-compose​

Start A-Parser:

docker compose up -d

Tuning threads in docker-compose​

When running via docker-compose, a limit on the number of open files/sockets applies inside the container. To increase the limit to 10240, add the ulimits block to the a-parser service:

services:
  a-parser:
    image: aparser/runtime:latest
    command: ./aparser
    restart: always
    volumes:
      - ./aparser:/app
    ports:
      - 9091:9091
    ulimits:
      nofile:
        soft: 10240
        hard: 10240

It is also recommended to perform Linux tuning for more threads on the host system.