2019 03 in Zipkin
Created by Adrian Cole on Apr 08, 2019
We began March with a welcome. Jorge Esteban Quilcate Otoya is now a committer and project team lead. Jorge also doubles as our Kafka expert and routinely presents at conferences. He's recently blogged about tracing and kafka, too! If you've not yet met Jorge, stop by gitteror send him a congrats tweet!
Next, we held a workshop at Salesforce. It focused on topics including data quality and the future of zipkin-js. Thanks very much to Chris for assembling things together, and introducing us to his teammates. Read Mapping into Zipkin data for notes on topics discussed. Right after the workshop, Medidata gave us a nice treat: their site doc now includes their Amazon bill! As users ask a lot about the cost of tracing, it is handy to have a point of reference.
Other good news is that zipkin4net, our .net library, is back in business. During our workshop, we discussed the dilemma that happens often in open source: the original authors of the project were no longer in a position to continue it. Thanks very much to Medidata for stepping up and re-owning this project. Specifically, Jordi drove things including fixing api keys to get a release out on Nuget. Thanks so much!
The middle of March focused on migrating more code to the Apache Software Foundation, specifically another repo move. brave-cassandra was our first Apache Incubating source release of code actually used in production. If you are interested in getting more Cassandra in your traces, give it a shot! We are planning to move the core zipkin repositories next.
Last, but not least, play-zipkin-tracing moved into our openzipkin-contrib org. Thanks very much to Shimamoto and Takezoefor continuing to support this in a place where non-employees can contribute!
Before we round up, let's chat on a few things we need community input for.
- Currently, we use Apache Spark to do "after the fact" processing of the service dependency graph. This means data are pulled from storage, usually like a batch job. Note, we also have projects for streaming aggregation, but after-the-fact is still default. Not everyone knows Spark well enough to support particularly cluster related troubleshooting. If you can contribute Spark support, please contact us!
- Adding a new default for moving trace data from apps to Zipkin is an important decision. We've had interest and code contributions for Apache ActiveMQ. If you would test and use this in production, please respond back to one of the issues listed on the related mail thread.
- Our Java instrumentation project, Brave, has a pull request awaiting votes for Alibaba Druid support. We need to see more interest before polishing this into the default build. If you would trace Alibaba Druid with Brave, please add your feedback here.
Here's a quick round of notable updates, broken down by org and repository.
ASF repositories:
brave-cassandra first ASF incubating release v0.10.2
OpenZipkin repositories:
brave fixes a number of subtle bugs
docker-zipkin is now using distroless JRE 11
zipkin uses Armeria for its http engine and includes a "Preview Lens UI" button
zipkin4net includes route filtering for Owin and AspNetCore middleware
zipkin-aws is developing a DynamoDB storage engine (our first SaaS-only storage)
zipkin-gcp Stackdriver storage now uses Armera implementation (instead of google grpc-netty library)
zipkin-go now runs against go 1.12 and gRPC 1.17
zipkin-js drops Node.js 6 support and adds got and axios instrumentation
zipkin-reporter-java removes redundant batching in the Kafka transport
zipkin-ruby supports v2 format and improves http status handling
OpenZipkin contrib repositories:
brave-ratpack fixes an trace scope issue working with promises
play-zipkin-tracing release includes Akka 2.5 and Play 2.7 support
Document generated by Confluence on Jun 18, 2019 18:50