- Upgraded from go 1.19 to 1.22
- Upgraded to Sarama v1.38.1
- Upgraded to Prometheus Go Client v1.14.0
- Upgraded to OpenTelemetry v0.13.0
- Fixed error-related metrics starting from 0 (#215)
- Add connection_total metric (deprecating connection_error_total metric which will be removed in future)
- Extended consumer latency recorded after partition leadership reverts to preferred (#204)
- Upgrade from go 1.13 to 1.19
- Use errors.Is when testing Sarama errors.
- Upgrade sarama to 1.37.2
- Added support for OpenTelemetry (#197)
- Only "jaeger" and "otlp" are supported as exporter protocols for tracing. See documentation for more details.
- Updated dependency to Sarama 1.35.0 (#200)
- Updated default Kafka version used by Sarama library to 3.2.0 (#200)
- Fixed replicas assignment taking rack configuration into account (#190)
- Use separate sarama clients for producer/consumer in order to reliably measure message latency (#188)
- Adjust default histogram bucket sizes (#194)
- Forced cleanup.policy to "delete" for the canary topic (#173)
- Allow to change canary logging verbosity and enable/disable Sarama logger at runtime (#166)
- Added Prometheus support via
PodMonitor
and sample Grafana dashboard for exposed metrics - Treat ETIMEDOUT (TCP keep-alive failure) as a disconnection condition too (#159)
- Updated dependency to Sarama 1.34.0 (#180)
- Updated default Kafka version used by Sarama library to 3.1.0
- Use 250ms as consumer fetch max wait timeout with Sarama 1.34.0 (#184)
- Added support for SCRAM-SHA-256 and SCRAM-SHA-512 authentication (#130)
- Added Python script tool to calculate latencies and quantiles from canary log for producer, consumer and connection buckets
- Turned off producer retry and added a consumer error metric (#133)
- Updated dependency to Sarama 1.30.0 with Kafka 3.0.0 support (#134)
- Fixed connection check service missing latency metrics for one broker (#141)
- Fixed wrong content type "text/plain" returned by status endpoint instead of "application/json" (#144)
- Fixed canary not recovering after a "connection reset by peer" from brokers (#148)
First implementation.