From 0913e550745872fe6d2f12d5e41320ba5899a1f3 Mon Sep 17 00:00:00 2001 From: "denis.bilenko" Date: Thu, 12 Nov 2026 14:23:48 +0000 Subject: [PATCH] acc: retry the spark-jar-task job task once when its cluster is lost The AWS test instance pool is spot-only in a single zone, so the job cluster is sometimes lost (spot kill, no capacity, terminated during library install). A lost cluster leaves no task logs, so the retried run prints the same output as a clean one. aws: TestAccept/bundle/deploy/spark-jar-task Co-authored-by: Isaac --- acceptance/bundle/deploy/spark-jar-task/databricks.yml.tmpl | 4 ++++ acceptance/bundle/deploy/spark-jar-task/test.toml | 3 +++ 2 files changed, 7 insertions(+) diff --git a/acceptance/bundle/deploy/spark-jar-task/databricks.yml.tmpl b/acceptance/bundle/deploy/spark-jar-task/databricks.yml.tmpl index 9078bcbdbbd..396ba5759b5 100644 --- a/acceptance/bundle/deploy/spark-jar-task/databricks.yml.tmpl +++ b/acceptance/bundle/deploy/spark-jar-task/databricks.yml.tmpl @@ -14,6 +14,10 @@ resources: name: "[${bundle.target}] Test Spark Jar Job $UNIQUE_NAME" tasks: - task_key: TestSparkJarTask + # The AWS test pool is spot-only in one zone, so the job cluster is sometimes lost + # (spot kill, no capacity, terminated during library install). A lost cluster + # produces no task logs, so a retried run prints the same output as a clean one. + max_retries: 1 new_cluster: num_workers: 1 spark_version: 16.4.x-scala2.12 diff --git a/acceptance/bundle/deploy/spark-jar-task/test.toml b/acceptance/bundle/deploy/spark-jar-task/test.toml index 6670a96c8c3..f225570f7c4 100644 --- a/acceptance/bundle/deploy/spark-jar-task/test.toml +++ b/acceptance/bundle/deploy/spark-jar-task/test.toml @@ -1,5 +1,8 @@ Cloud = true +# Room for one task retry, which waits for a new job cluster. +TimeoutCloud = '40m' + Ignore = [ 'myjar/PrintArgs.jar', 'myjar/PrintArgs.class',