Open pavolloffay opened 3 years ago
Here are log from the query service 44s:
Picked up JAVA_TOOL_OPTIONS: -XX:InitialRAMPercentage=50.0 -XX:MaxRAMPercentage=75.0 -javaagent:/mnt/hypertrace/hypertrace-agent-all.jar
[main] INFO org.hypertrace.agent.core.config.HypertraceConfig - Config loaded: {"serviceName":"query-service","reporting":{"endpoint":"http://collector.traceableai:9411/api/v2/spans","opa":{"endpoint":"http://opa.traceableai:8181/","pollPeriodSeconds":30}},"dataCapture":{"httpHeaders":{"request":true,"response":true},"httpBody":{"request":true,"response":true},"rpcMetadata":{"request":true,"response":true},"rpcBody":{"request":true,"response":true},"bodyMaxSizeBytes":131072},"propagationFormats":["TRACECONTEXT"]}
[opentelemetry.auto.trace 2021-01-08 17:38:23:811 +0000] [main] INFO org.hypertrace.agent.core.config.HypertraceConfig - Config loaded: {"serviceName":"query-service","reporting":{"endpoint":"http://collector.traceableai:9411/api/v2/spans","opa":{"endpoint":"http://opa.traceableai:8181/","pollPeriodSeconds":30}},"dataCapture":{"httpHeaders":{"request":true,"response":true},"httpBody":{"request":true,"response":true},"rpcMetadata":{"request":true,"response":true},"rpcBody":{"request":true,"response":true},"bodyMaxSizeBytes":131072},"propagationFormats":["TRACECONTEXT"]}
[opentelemetry.auto.trace 2021-01-08 17:39:07:507 +0000] [main] INFO io.opentelemetry.javaagent.tooling.OpenTelemetryInstaller - Installed span exporter: io.opentelemetry.exporter.zipkin.ZipkinSpanExporter
[opentelemetry.auto.trace 2021-01-08 17:39:07:908 +0000] [main] INFO io.opentelemetry.javaagent.tooling.PropagatorsInitializer - Added tracecontext propagator
Hypertrace agent started, version: 0.7.5
First, we should see if there is any regression between the upstream build and HT and then try to figure out how to improve the startup time.
one of the possible reason could be cpu throttling. in case of api-anomaly-detector, it has cpu request=0.25 and limit=0.5. How much cpu does it need?
How much cpu does it need?
There is no minimum required CPU, however as you pointed out restricted resources can affect startup time.
Startup in api-anomaly-detector service in Traceable takes around 30s at least based on the agent logs