Support for invite user flow and multi-node high availability in Kubernetes

Any thoughts on supporting invite user flow? This forked repo of casdoor did it PR.

can you give more details about this feature?

What's the user scenario?

admin sends email granting access to a user to signup in an organization. Also, as the open source casdoor is designed to only work in a single node setup, are there any plans to make it highly available in kubernetes where it can scale without the enterprise solution?

Best choice: inspect and rewrite the code to feed your needs

will the user need to create his account with that email? Or other emails also work?

casdoor supports deployment via k8s helm, see: https://casdoor.org/docs/basic/try-with-helm/

I know casdoor supports kubernetes as I'm the one who deployed it to dockerhub that the docs reference.

There seems to mixed messages about high availability about this project. On one hand, the doc u linked states "scalable management" with the chart on kubernetes; however, on the other hand, you have stated there is no multi node support for open source which is a strict requirement for high availability (see the linked thread this message replied to). In the same linked thread, you mentioned the enterprise solution is stateless. Is open source solution stateless or only enterprise is? It's important to know if this open source can scale in production.

It would be limited to same email that was invited

casdoor has everything you need to build a multi-node deployment. You can put frontend files into a cloud CDN, you can add a LB before the backend APIs. The session can be shared via Redis. So there's nothing stopping a guy from deploying casdoor to multiple nodes

Of course this project itself doesn't provide a out-of-the-box clustering management system to manage a casdoor cluster. I don't think most people need that as SSO is actually a low-traffic service and usually one node already fulfills 99% needs. They must be very large companies like FLAG that will need multi-node deployment of Casdoor. Those big companies either choose to build their cluster management system by themselves or choose an enterprise backed solution.

I mean, most open source projects won't aim at serving 1,000+ millions users or data size at first (except very popular and widely used big data projects like Hadoop). as most of the players only have scale like 1,000 users. So the focus of the open-source project is providing the best quality software for the reasonable size of users or data size, not the size like FLAG.

Of course, if the current Casdoor code has created a barrier that stopped anyone from building a multi-node cluster, and it can be fixed with very straightforward PRs, it's also welcome to propose such PRs

invitation code pages already added: https://github.com/casbin/casdoor/commit/2fb79e4092f5a4094070f26ca92711f10129a47c