andygrove opened a new issue, #6389:
URL: https://github.com/apache/datafusion-comet/issues/6389

   ### What is the problem the feature request solves?
   
   A third of the failed merge-queue runs between 2026-09-11 and 2026-09-29 (11 
of 32) failed only on a network fetch. Each one removed a pull request from the 
queue, and it had to be re-queued by hand at the back of the line. Seven of the 
eleven were in the Spark SQL test jobs:
   
   - the sbt launcher fetching sbt itself, two to four seconds into `Run Spark 
tests`: `[error] [launcher] could not retrieve sbt 1.11.7` (3 jobs)
   - sbt resolving its own dependencies or the build's plugins from Maven 
Central: `Connection reset` while downloading `gigahorse-apache-http`, 
`maven-resolver-transport-file`, `jarjar-abrams-core` and 
`librarymanagement-ivy` (4 jobs)
   
   The build job's `Pre-compile Spark Test classes` step already retries 
resolution failures (#6078), but each of the test jobs starts its own sbt, 
downloads the same things again, and does not retry. The Iceberg jobs have the 
same gap on the Gradle wrapper download: `SSLHandshakeException: PKIX path 
building failed` in shard 4 on 2026-09-23, and a `Connection reset by peer` 
variant seen before.
   
   ### Describe the potential solution
   
   - In `spark_sql_test_reusable.yml`, retry `Run Spark tests` up to twice when 
sbt fails on a download before any test has run, meaning the log has a 
resolution error and no test result lines. A failure after tests start is never 
retried.
   - Add `could not retrieve sbt` and `Error downloading` to the patterns the 
pre-compile step retries on.
   - In `iceberg_spark_test_reusable.yml`, install the Gradle distribution in 
its own step (`./gradlew --version`) with retries, before the steps that run 
the tests.
   
   ### Additional context
   
   The other network failures in that window (a `setup-java` DNS failure, an 
artifact `ECONNRESET`, a Maven 429 in the PyArrow job) were one-offs and are 
left alone.
   


-- 
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]

Reply via email to