hydrus server running on Debian
Unmaintained: this project is no longer maintained. I suggest you check out suika/hydrus as an alternative (Docker images for both hydrus client and server are provided).
This is a simple Debian-based Docker setup for running hydrus server from source.
The latest build runs hydrus server version 430.
The easiest way to install is via Docker Hub:
user@local:~$ docker pull mtbl/hydrus-server-dockerBy default, this will pull the latest build. To specify an image with a specific version of hydrus server, provide the version number as tag, e.g.:
user@local:~$ docker pull mtbl/hydrus-server-docker:430Alternatively, you can also build the image yourself:
user@local:~$ git clone --recurse-submodules https://github.com/imtbl/hydrus-server-docker.git
user@local:~$ cd hydrus-server-docker
user@local:hydrus-server-docker$ docker build . -t hydrus-server-dockerThe user that is used inside the container has UID 1000 and GID 1000 by
default. You can adjust this (e.g., to match your host UID/GID) by providing
the arguments USER_ID and GROUP_ID when making a build.
If you have installed via Docker Hub, just pull the updated image. Otherwise, pull from this repository and make a new build.
This repository follows semantic versioning and any
breaking changes that require additional attention will be released under a new
major version (e.g., 2.0.0). Minor version updates (e.g., 1.1.0 or 1.2.0)
are therefore always safe to simply install via the routine mentioned before.
When necessary, this section will be expanded with upgrade guides to new major versions.
Upgrading from 3.x to 4.x can be done via pulling the updated image from
Docker Hub or building it yourself and requires no further manual changes.
4.0.0 has introduced no breaking changes and merely reflects the switch to a
new license (AGPLv3).
Upgrading from 2.x to 3.x can be done via pulling the updated image from
Docker Hub or building it yourself and requires no further manual changes.
Due to the introduction of OpenCV to hydrus server (which is hard to build on Alpine), the Docker image is now based on Debian instead of Alpine.
Upgrading from 1.x to 2.x can be done via pulling the updated image from
Docker Hub or building it yourself and requires no further manual changes.
Since there have been issues with running the pre-compiled version of hydrus
server 335+ (which made the switch to Python 3) on the previous Docker setup,
starting with 2.0.0, the Docker image is now based on Alpine (instead of
Debian) and runs hydrus server from source.
This approximately halves the resulting image size while at the same time making it easier to adapt for future changes/dependencies.
First, you need to bind the exposed ports. This can be done automatically using
-P but it is recommended to bind them manually instead since having changing
ports every time you run a new container might be annoying when used in
combination with other services.
hydrus-server-docker exposes the following three ports by default:
45870/tcp45871/tcp45872/tcp
These are used to access the different services. 45870 is the default port
for the server administration service while 45871 and 45872 are used for
repositories. You will generally have two (one for tags and one for files), but
if you add more, you will also need to expose additional ports.
Per default, hydrus-server-docker stores its databases and media inside the
/data directory which is a mount point that is persisted as a volume. A new
volume will be created every time a container is created, making it less ideal
as a long-term solution. Instead, you should create a named volume yourself and
mount that over it:
user@local:~$ docker volume create hydrus-server-dataAfter creating your named volume, you can run the container. Here is a full example with all the options mentioned above:
user@local:~$ docker run -p 45870:45870 -p 45871:45871 -p 45872:45872 -v hydrus-server-data:/data -d mtbl/hydrus-server-dockerSpecifying the same named volume every time a container is created gives each of these instances access to the same persisted data.
Of course, using a bind mount instead of a named volume is also possible but for performance reasons only recommended if you need easy access to the data on the host machine.
You are welcome to help out!
Open an issue or submit a pull request.
AGPLv3 © imtbl