cxzl25 commented on code in PR #3488:
URL: https://github.com/apache/celeborn/pull/3488#discussion_r2375190516
##########
common/src/main/scala/org/apache/celeborn/common/CelebornConf.scala:
##########
@@ -3915,6 +3917,17 @@ object CelebornConf extends Logging {
.booleanConf
.createWithDefault(false)
+ val WORKER_GRACEFUL_SHUTDOWN_DB_DELETE_FAILURE_POLICY: ConfigEntry[String] =
+ buildConf("celeborn.worker.graceful.shutdown.dbDeleteFailurePolicy")
+ .categories("worker")
+ .doc("Policy for handling DB delete failures during graceful shutdown. "
+
+ "THROW: throw exception, EXIT: trigger graceful shutdown, IGNORE: log
error and continue (default).")
+ .version("0.7.0")
+ .stringConf
+ .transform(_.toUpperCase(Locale.ROOT))
+ .checkValues(Set("THROW", "EXIT", "IGNORE"))
Review Comment:
Since RocksDB cannot self-recover after encountering an exception, exiting
(via `EXIT`) allows the worker to shut down gracefully.
In the case of `IGNORE`, there will be no issues with cleaning up and
releasing disk space. However, this may affect the next restart of the worker,
as no new state will be written during recovery.
--
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]
