Closed AlexHockey closed 3 years ago
We forgot to release v1.8.0 to avoid this webrick issue.
https://github.com/fluent/fluent-plugin-prometheus#requirements
v1.8.0 and async-http
gem combo fixes this problem.
Thanks very much, I'll give that a go.
It appears that the fluentd only allows 5 concurrent connections to the metrics endpoint. Prometheus uses persistent connections, so this means that a single fluentd instance can only be monitored by 5 Prometheus instances.
This is unlikely to be an issue in production, but I just ran into it on a test system and it caught me by surprise. Is it worth documenting this somewhere, or producing a log when this happens?