Running WebHare in Docker


The recommended way to run WebHare inside a docker for development is:

  docker run -ti --rm --name webhare -p 80:80 -p 443:443 \
  -v ~/whdata:/opt/whdata webhare/webhare-core:master

On macOS and Linux, ~ refers to your home directory. On Windows you should replace ~ with %USERPROFILE% and type the entire command on one line, removing any \ characters at the end of the lines.

What these options do:

Using a data volume

Using a data volume is often more stable and improves performance on Windows and macOS hosts, but requires you to take care to not accidentally delete it as it will live inside your Docker virtual machine. If you go this route, we recommend frequently backupping it.

When using a data volume you may still want to mount /opt/whmodules to a normal path and use that to store your module development.

To create and use a volume, and have a separate module dir in whmodules:

  docker volume create webhare-data
  docker run -ti --rm --name webhare -p 80:80 -p 443:443 \
  -v webhare-data:/opt/whdata -v ~/whmodules:/opt/whmodules webhare/webhare-core:master

For more information see Manage data in Docker

Docker caveats

If you run Docker inside a VirtualBox VM, do not store WebHare's data or modules folder on a VirtualBox 'shared folder'. WebHare requires the use of mmap() which will fail on this filesystem. See also []


Manual startup

By appending /bin/bash to a docker started with -ti you can manually control the startup. Once in the container, run /opt/container/ to start WebHare.

You can also wh exec postgres to start just the postgres database and then use wh psql to connect to this database if you need to fix pre-launch issues.

If you cannot startup the database server, use wh postgres-single to launch the database in single user mode. You'll land in a command prompt immediately.

Core dumps

Process core dumps are disabled by default but you can enable them with prlimit:

docker exec -ti <containerid> /bin/bash
prlimit -p `pidof webserver` -cunlimited:unlimited

You can force a coredump by sending a QUIT signal to the process (eg pkill -QUIT webserver).

Coredumps generally end up in the /tmp/ directory. If you cannot find the coredumps, check the various /proc/ settings.