Closed midnightconman closed 4 weeks ago
This is an interesting topic. I'm going to do this in my Druid cluster to improve observability.
In contrast to provide such ability by Druid itself, it's highly recommended to use some non-intrusive(auto instrumentation) solutions such as PinPoint to achieve this goal. As I know, pinpoint works very well with Jetty used by Druid. If I understand correctly, open telemetry also provides auto instrumentation.
So the only thing we have to do is putting java agent of those tools mentioned above to the java command line of Druid services. But one thing that I think we need to make little change is at the peon command line. It's better to set different application name, which will be reflected at the observability platform, for each task based on their datasource so that it's more clear that a query goes into which task.
Any update on this feature?
This issue has been marked as stale due to 280 days of inactivity. It will be closed in 4 weeks if no further activity occurs. If this issue is still relevant, please simply write any comment. Even if closed, you can still revive the issue at any time or discuss it on the dev@druid.apache.org list. Thank you for your contributions.
This issue has been closed due to lack of activity. If you think that is incorrect, or the issue requires additional review, you can revive the issue at any time.
Description
How does the apache druid project feel about tracing support in Druid? As tracing gains more traction in the industry, more and more applications are relying on it to provide insight into the timings of the various functions and how long various activities take.
Opentelemetry is a pretty popular tracing library (with java) support these days : https://github.com/open-telemetry/opentelemetry-java
Motivation
The ability to send tracing data from druid components to a tracing provider / backend. This would help in tracking and improving function timing for druid.