A super simple dynamic proxy for rootless podman development.
Run it:
pipenv run python podproxy.py <YOUR_HOST_IP> [<HOST_PORT>]
Podman proxy will do the following:
- At startup it will go through the running containers and create virtual host configs for them.
- After startup, it will listen to podman events and update the virtual host configs accordingly.
- It will reload nginx when ever configs change
When you start your podman containers, specify a hostname and expose a port.
Podman proxy will config a virtual host with the specified hostname and reverse
proxy to the exposed port. You can specify the internal port without specifying
the external port to let podman chose the port -p 80. Podman proxy will grab
the random port from the container.
If you use subdomains of localhost (ie: test.localhost), you don't need to do
anything else, otherwise something like dnsmasq will be needed which is not
covered by this tool.
Podproxy will create a network named 'podproxy'. Any containers that use the
'podproxy' network and has a port defined in it's config (PORT specified in
Dockerfile), then podproxy will proxy to that container through the podproxy
network. This allows you to create containers without exposing ports on the
host. Once again, hostname has to be defined on the container for podproxy to
proxy to the container.
Podproxy will try to run on port 80 unless another port is specified as the 3rd
second argument after YOUR_HOST_IP. Normally podman doesn't have permission to
ports lower than ???. If using the specified port fails, podproxy will start on
a random port.
Podproxy currently does not support https as for most cases, http will work for development purpose. But for those rare cases that https is needed for development, https support is planned for the future.
2 test compose.yml folders has been added for testing purposes.
When you create a podman container, you can specify a hostname for the container
using --hostname flag. If using compose file, specify hostname in the
service. Podproxy will then proxy requests to that hostname, to that container.
For example, if you start a container with hostname app1.localhost, then you
can make a request to http://app1.localhost and podproxy will proxy that request
to the exposed/specified port on that container. if you use *.localhost it
will work out of the box. But if you hostname is something like app1.dev then
you will need to tell you system to resolve app1.dev to 127.0.0.1 (or any
loopback address). You can do this using /etc/hosts file:
sudo sh -c "echo '127.0.0.1 app1.dev' >> /etc/hosts"
Or you can use dnsmasq to resolve any *.dev domain to localhost (probably) but
I haven't checked on this.
- Currently if multiple ports are exposed, only the first port is proxied to.
- If multiple containers run in a pod without networking, only one port from this pod will work.
- Create a proper package
- Create a systemd user unit
- Separate nginx related stuff to a separate module
- Expose a web dashboard on the default host and show current status
- Find a better way to run on port 80 without sysctl configuration
- Integrate with Netavark