Open bdbene opened 2 years ago
If the code were a tag, you can use regex promql matches on ^4..
and ^5..
+1. @rayokota wonder if your team could help with this request?
@Claimundefine wonder if your team could help with this request?
To conclude after looking into the code, SR jmx actually has the http_status_code
needed for the purpose from the underlying jersey server. @bdbene @OneCricketeer
The JMX metric that currently tracks errors doesn't differentiate between client errors (400s) and server side errors (500).
jmx.kafka.schema.registry.request_error_rate
It would be very helpful from an operational point of view to be able to distinguish between those types of errors. For example if there were two metrics along the lines of:
Then anyone maintaining a Schema Registry instance would be able to set up monitoring and alerting on schema-registry errors, and not have to worry about false alarms every time a client uses the service incorrectly.