Hi! I spent the past week working on HBASE-22120 <https://issues.apache.org/jira/browse/HBASE-22120> (Replace HTrace with OpenTracing). I planned to do this awhile back and am pleased to finally carved out some time to dedicate to this project.
Here's a scope doc that describes what I plan to do and milestones. https://docs.google.com/document/d/1Q1zQKzxxymE4vcOBYElkQopmh2QvvYRBlnDqMEBPsB0/edit?usp=sharing I now have a proof of concept that: (1) replaced every Java HTrace code with OpenTracing/JaegerTracing. (2) fixed the trace propagation bug introduced in HBASE-18601 when it attempted to update from HTrace 3 to HTrace 4. (2) TestOpenTracingHooks (which is modified from TestHTraceHooks) passes, so that means at the very least, create table operation works. Tasks remain to be done: (1) more tests to make sure other existing traces works (2) migrate HBase shell (in Ruby) and scripts to use OpenTracing (3) update docs (4) make it easier to support other OpenTracing tracers. (5) propagate the traces to other systems such as HDFS and MapReduce. Now, this work is going to span a few jiras and it touches pretty much every part of HBase. Would it make sense to do it in a feature branch? Thoughts? Weichiu
