Serveo.net is down – Causes CI-tests to fail

[LogFlames]

When trying the CI-tests here 04b8ea3 i noticed that serveo.net is currently down, which causes the entire CI workflow to fail as well.

We could either choose that to be the desired behavior, or set the APPLICATION_URL=http://localhost:8000 and trigger the tests with an additional --exclude-tag=external in case serveo.net is not reachable. We would however not see that the swish-tests didn't run.

If you want to have the fallback of only running local tests, I will be happy to add that to the workflow.

[LogFlames]

Serveo.net is up again, if it's a one time occurance that it goes down we are probably fine without any special treatment

[JonathanGodar]

Not a onetime thing. Happens quite often. localhost.run is more reliable but you cannot get a custom subdomain so the domain name would have to be extracted from the ssh command. Furthermore the hostnames change regularly on localhost.run, even without restrating the ssh command. However, given that the tests should not run for that long, localhost.run should not have the time to change the url.

Summary: Would be nice if we could get a ssh forwarding service on fysikmotorn, otherwise localhost.run is better for tests i think and serveo is better for development (when it is working)

[Gabrimari]

As of 927073a we now use localhost.run for CI tests instead of serveo since that service seems to be more reliable. I will leave this issue open for a while to see if localhost.run is reliable enough for the service. Otherwise we will have to look into alternative solutions. Either creating our own service or some fallback option.

[LogFlames]

We will try to setup a service on fysikmotorn, should be quite easy to create it using a static port, but want to explore how to allocate the local port on fysikmotorn to allow simultaneus connections nicely.