chore: log if metrics_endpoint but no feature flag #5279
Conversation
|
Nice, now that @wileyj updated the test runner to use the |
|
simple_neon_integration test seems to be broken and I think by this...it is complaining I think about a metrics response not existing as expected. |
| @@ -1500,19 +1500,16 @@ fn simple_neon_integration() { | |||
| // query for prometheus metrics | |||
| #[cfg(feature = "monitoring_prom")] | |||
| { | |||
| wait_for(10, || { | |||
There was a problem hiding this comment.
Pretty sure this wait_for was there for a reason. Maybe it's because the thread that runs the metrics hasn't started by the time this function is called?
There was a problem hiding this comment.
This applies to all the other wait_fors below, as well.
There was a problem hiding this comment.
Yeah, I don't remember my exact thought process, but Jacinta recently had added the wait_fors when fixing these in a different PR. I may have just removed them from a bad merge. I've added them back.
It's not that the metrics server hasn't started, but the metric itself isn't updated until somewhere down the relayer stack, and I believe that's why the metric isn't updated "immediately".
When calling
start_server_monitoring_metrics, there is an inverse feature flag (not(features = "monitoring_prom")) and log if you've setmetrics_endpointbut don't have--feature monitoring_prom. This is helpful in the case that you're trying to setup monitoring, and added the config field, but didn't add the feature flag.This log wasn't getting shown because
start_server_monitoring_metricswas itself behind a feature flag, so the log never showed up. This PR changes this to always callstart_server_monitoring_metrics- it's still a no-op (with just the log mentioned) if the feature flag isn't provided.