Additional info about EVE state (FEATURE REQUEST)

[sadov]

It seems reasonable to consider the possibility of obtaining more complete information about the state of EVE, may be by request. At this time, we didn’t have any information about EVE network instances, volumes, and general EVE uptime in existed periodic info/metrics. And the logs of the state changes of these things can be lost during of something failures to save the logs in the controller part. Such a solution can help implement more stable controller-side EVE models, especially if we have a stateless controller.

[rvs]

It seems reasonable to consider the possibility of obtaining more complete information about the state of EVE, may be by request.

This has been discussed a number of times, but the devil is in the details -- please provide a list of things you'd like EVE to tell controller

At this time, we didn’t have any information about EVE network instances, volumes, and general EVE uptime in existed periodic info/metrics.

I find this hard to believe -- there's plenty of network instances info/metric stuff -- what am I missing?

And the logs of the state changes of these things can be lost during of something failures to save the logs in the controller part.

The logs can be delayed, but unless the buffer overflows (which would take quite a bit of time) they are guaranteed to be delivered to controller eventually. Are you seeing something else?

especially if we have a stateless controller.

And therein lies the rub -- controller can't really be stateless. I mean, you can get away with some things while still thinking of controller as stateless, but at some point you'll have to bite the bullet and implement state there.

In fact, this is how commercial controller is organized. It may be time to organize Adam the same way -- but I need more information to be fully convinced.

[mydatascience]

https://lists.lfedge.org/g/eve-tsc/topic/79183030#636

[temerkhanov]

We should take into account a completely valid use case when we can have multiple USB controllers loading some firmware. In this case we can have endpoint addresses differing between runs.