Question and Answer count

Thanks @user620588. The number of questions that are asked versus the number of questions are answered may indeed make a metric to determine the site health.

That graph shows a steady decline in the absolute number of questions asked and answered on ServerFault.

So by that metric the site is in decline and has been for years.

2020 saw a trend reversal, but only in the number of questions asked rising sharply.
(Maybe because people working from home due to restrictions get less mentoring and/or are less inclined to "bother" their colleagues to ask a question and turn to online resources instead?)

The absolute number of answered questions in 2020 didn't show any increase at all, but rather continued the downward trend. So percentage-wise much more questions remain unanswered now. That aggravates the appearance of decline of the site.

Those numbers suggest more sand and fewer pearls...
https://stackoverflow.blog/2011/06/13/optimizing-for-pearls-not-sand/

Voting

Since you need to be registered user to be able to vote the number of votes cast might be another metric to show “community” and participation :

https://data.stackexchange.com/serverfault/query/1375382/up-accept-and-down-votes-over-time#graph