[Bug] Theia cloud in https://try.theia-cloud.io/ is not accessible.

[inlann]

When I tried to launch the Theia Blueprint, the page shows an error:

Oh no, something went wrong! ☹
ERROR: Failed to execute 'replace' on 'Location': 'https://' is not a valid URL.

[inlann]

BTW, I have another question about the deployment of Theia Cloud.

I deployed Theia Cloud in my Kubernetes cluster and noticed that the Operator module creates an Ingress for each user's IDE Pod, resulting in access URLs like 192.168.2.152.nip.io/instance/ for each IDE Pod. However, when I tried the official demo at https://try.theia-cloud.io/, I noticed that the URL for my IDE Pod did not have the /instance/ format but rather appeared as xxxx.instance.theia-cloud.io, using a subdomain format.

I observed that my DNS service provider supports wildcard domain configurations like *.test.example.com. I would like to know if the official Theia Cloud setup uses this method, and how I can configure my Theia Cloud deployment to use subdomains in this way.

[sgraband]

Like i said it will definately be possible, but you will need to do customizations to the code.

[inlann]

Thanks, I'll try it.

BTW, can I use my own ssl certificates rather than the cert-manager?

[jfaltermeier]

With the 0.11 release that is planed for July (or the current main branches), there will be these helm values:
ingress.certManagerAnnotations: bool
ingress.addTLSSecretName: bool
https://github.com/eclipsesource/theia-cloud-helm/tree/main/charts/theia-cloud

You can set ingress.certManagerAnnotations to false and ingress.addTLSSecretName to true. Then you may provide the certs via the secret names like this:
https://github.com/eclipsesource/theia-cloud-website/pull/32/files#diff-657dc246a5ef715c9d6c7eaa8bd383c66d2e463855e6b9b951f1c806aa3c0402R128-R132
(this is WIP documentation that will be published with the 0.11 release)

You may also set ingress.addTLSSecretName to false and patch the ingress controller to use a default certificate that will work for your use case.

[inlann]

Like i said it will definately be possible, but you will need to do customizations to the code.

I tested Gitpod, and it creates a subdomain for each user's pod. However, I found that CDN service providers can configure origin URL rewriting. They can use regex matching to remove the /instances/<uuid> string from /instances/<uuid>/xxx.js. Our current strategy is to modify the static resource URLs to a fixed format like cdn.example-domain.com/xxx.js when packaging the source code and to provide a dedicated static resource server to store these static resource files. Therefore, placing the UUID in the path will not affect our CDN acceleration.

[inlann]

You may also set ingress.addTLSSecretName to false and patch the ingress controller to use a default certificate that will work for your use case.

That sounds great! Looking forward to the release of version 0.11.

[sgraband]

The error on the try-now page should be resolved. Thank you for reporting. We will improve the error messages with our next release.

[7ojo]

Still seeing same error. Does not work for me on MacOS + Firefox

[sgraband]

Thanks for reporting. We will check 👍

[sgraband]

We are experiencing unusual load at the moment and the cluster is simply maxed out. We will monitor this and provide a fix asap so that the correct error message is displayed. Thank you for your patience and reporting this.