@agile6v We can not to use etcd to store the node status, it will cause the data version in etcd to keep changing.

Hi @membphis

I realized that etcd is really not suitable for storing data that change frequently.
Do you have any good idea?

Thanks.

how about adding these data in the Prometheus plugin? Or send to elk?

I don't think it's a good choice to introduce a heavy component. I want to split this feature and first implement the /apisix/status api enhancement. What do you think of?
@moonming

how about reusing the API in the Prometheus plugin: https://github.com/apache/incubator-apisix/blob/master/lua/apisix/plugins/prometheus.lua#L52? This way will not introduce Prometheus, you can use any component to call this API.

I realized that etcd is really not suitable for storing data that change frequently.
Do you have any good idea?

The prometheus is better.

@agile6v what's your opinion?

@agile6v what's your opinion?

@moonming I haven't figured out how to display the cluster information. This plugin can display the information of the single machine. But Is the information of the cluster read from prometheus?

@agile6v Of course, it can be implemented in prometheus.

@agile6v Of course, it can be implemented in prometheus.

OK. I will implement it ASAP.

@agile6v do you want to continue work on this issue or we can close this PR first?

OK. Recently learned about the implementation of k8s-ingress. It should be what we want.