Yang Jie created SPARK-58674:
--------------------------------
Summary: Assign a name to the error condition
_LEGACY_ERROR_TEMP_3021-3029
Key: SPARK-58674
URL: https://issues.apache.org/jira/browse/SPARK-58674
Project: Spark
Issue Type: Sub-task
Components: Spark Core
Affects Versions: 5.0.0
Reporter: Yang Jie
Convert five remaining {{_LEGACY_ERROR_TEMP_*}} conditions in
{{SparkCoreErrors}} to proper error conditions, continuing the cleanup under
SPARK-37935.
Four of them get user-facing names, and one is judged an unreachable defensive
check and becomes an internal error:
* {{_LEGACY_ERROR_TEMP_3021}} -> {{SCHEDULER_BACKEND_SHUTDOWN_FAILED.EXECUTORS}}
* {{_LEGACY_ERROR_TEMP_3022}} ->
{{SCHEDULER_BACKEND_SHUTDOWN_FAILED.DRIVER_ENDPOINT}} (new umbrella, SQLSTATE
58030)
* {{_LEGACY_ERROR_TEMP_3023}} -> {{SparkException.internalError}}, JSON entry
deleted
* {{_LEGACY_ERROR_TEMP_3026}} -> {{UNSUPPORTED_CALL.TASK_NOT_FINISHED}} (new
subclass of an existing umbrella, SQLSTATE 0A000)
* {{_LEGACY_ERROR_TEMP_3029}} -> {{CLUSTER_MANAGER_APPLICATION_FAILURE}} (new
top-level, SQLSTATE 56000)
{{_3023}} guards {{DAGScheduler.submitMapStage}} against a zero-partition RDD.
Its only production caller, {{ShuffleExchangeExec.mapOutputStatisticsFuture}},
already short-circuits that case, and both {{DAGScheduler}} and
{{SparkContext.submitMapStage}} are internal APIs, so the branch is only
reachable by violating the internal-API contract. Per the error-conditions
README that makes it an internal error rather than a named user-facing
condition.
The remaining conditions in the {{_3021-3042}} range ({{_3028}}, {{_3033}},
{{_3035}}, {{_3036}}, {{_3037}}) are left for follow-up work.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
