NoahKusaba commented on PR #2478:
URL:
https://github.com/apache/datafusion-ballista/pull/2478#issuecomment-5786530252
Besides the two review fixes, I went back through the whole PR checking each
claim against the source, and pushed a second commit (bc54b880) with what
turned up:
- **Spark functions from the CLI.** I had written that `ballista-cli` picks
up Spark functions from the cluster. That was wrong: SQL is planned in the
client, so a CLI without them rejects `sha1(...)` before anything reaches the
cluster (the same reason `remote-spark-functions` requires
`ballista-core/spark-compat`). The guides now say a `cargo install`ed CLI can't
call them, and give the source build that can: `cargo build -p ballista-cli -p
ballista-core --features ballista-core/spark-compat --release`.
- **Sample logs.** I'd only bumped the version numbers in the Docker,
Compose and Kubernetes samples, but the messages themselves were stale too. The
executor sample also showed pull-mode output, although push-staged is the
default. They're now re-captured from a locally built scheduler and executor.
- **Benchmark formats.** I had switched two tpch commands to `--format
parquet` for no reason; `tbl` works, so they're back to `tbl`.
- **Smaller notes.**
- The KEDA section now says the published images are built without
`keda-scaler`.
- The REST API section mentions `--disable-rest-api`.
- The Prometheus note is tidied.
- The plan-stability README no longer quotes `rust_clippy.sh`'s exact
command, since #2473 changes it.
The description is updated to match.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
