implement running and total mnesia node count metrics

[luos]

Hi,

It would be useful for us to expose running and total mnesia node counts.
Let me know what you think.

Thanks

[deadtrickster]

Hi, I'm somewhat concerned with arity here. can peer label explode?

[luos]

Hi,

I guess, that is a a good question. In our typical installation it's about a maximum of 7 nodes connected in an mnesia cluster. It should be same or less than the node_state metric coming from the dist collector (if all nodes online).

I am not sure if others use many more mnesia nodes?

If you have many nodes, let's say the mentioned 7, that would create 7*6 metrics in total, so I guess you can get somewhat flooded by these. Though if you already have 7 nodes added, an additional 6 metrics for each should not be a problem.

Another case I thought of can be if people have dynamic nodes in a cluster and do not clean up the removed/shutdown nodes properly, but I guess then that is more of a bug than intended.

[luos]

Hi,

Do you think this can be merged, should I change something? Do you still have concerns?

Thanks