Originally discussed in #4391 for datasource timeouts, but the problem is more broadly applicable to all health indicators.
Originally discussed in #4391 for datasource timeouts, but the problem is more broadly applicable to all health indicators.