These past 2 weeks we have been focused on wrapping up the zcashd deprecation work and released Zebra v3.0.0 and v3.1.0.
New features for Zebra include a dust filter that rejects transactions with very small outputs from the mempool, helping protect the network from certain types of spam. The RPC response size limit was increased, allowing larger responses for applications that need them. Several fixes improved stability: a panic in the async transaction verifier was addressed, backtraces were enabled in production builds to help with debugging crash reports, and a logging issue was fixed for users installing Zebra via cargo. The documentation tooling was also updated to fix book rendering.
Several improvements were made to Zebra’s Docker infrastructure during this period. The base Docker image was updated from Debian Bookworm to Trixie. Support for both ARM64 and AMD64 architectures was properly implemented by fixing the multi-arch Docker manifest creation, ensuring users on different hardware can pull the correct image. A fix was also applied to correct commit ID logging in Docker images.
The CI pipelines received multiple fixes to improve build reliability. Docker builds were failing due to disk space limitations on standard GitHub runners, which was resolved by adding configurable runner types that allow larger runners to be used when needed. An authentication issue with Docker Scout on release PRs was fixed, and a problem with static IP assignment in deployment workflows was corrected. On the deployment side, GCP instances deployed from main and releases now have RPC enabled), making it easier for external services to connect to these nodes.
Outside of Zebra we have been supporting the work on ZSA test vectors modifications.
For December we are going to be having a break from our usual programming to run a Hackmas and work on things of our choosing, so expect a smorgasbord of features and improvements in our next update.
Thanks for reading!