Can we create a new APM service from a subset of APIs which represent a common business logic, as check logs in one big APM service doesn't make sense
Service 1 has multiple child services. Services 5 to 10 are a subset of service_1 which service_1 uses anytime. Can we have separate APM service_name for Service 1 and Services 5 to 10.? So a team can view only their own transactions and not get lost in distributed tracings spans.
I think your best bet may be to set labels for different parts of your app using elasticapm.label() -- then, you can filter by them in the UI. But they still won't show up as separate services, unfortunately.
In v5.10.0 we introduced the ability to replace the global excepthook with an Elastic handler. If you set USE_ELASTIC_EXCEPTHOOK=True you can enable this feature and that should help plug some of the holes in your exception reporting.
Note that if a framework catches uncaught exceptions automatically, this feature may not solve your problem. In that case you need to hook into the framework's exception handling and capture_exception there.
okay, I think I have a newer problem on my side, so we have log handlers and use filebeat to send logs, which works perfectly fine. Now, after adding elastic APM agent, all requests which have 200 stat are shown as errors.
That sounds very odd. Anything in the application/agent logs?
You might also consider checking out ecs-logging-python which provides ECS-compliant formatters for easy ingestion via Filebeat, including log correlation with APM traces.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.