stevenzwu commented on code in PR #4687:
URL: https://github.com/apache/iceberg/pull/4687#discussion_r864410262
##########
spark/v3.2/spark/src/main/java/org/apache/iceberg/spark/source/SparkWrite.java:
##########
@@ -196,24 +199,33 @@ private void commitOperation(SnapshotUpdate<?> operation,
String description) {
}
long start = System.currentTimeMillis();
- operation.commit(); // abort is automatically called if this fails
- long duration = System.currentTimeMillis() - start;
- LOG.info("Committed in {} ms", duration);
+ try {
+ operation.commit(); // abort is automatically called if this fails
+ long duration = System.currentTimeMillis() - start;
+ LOG.info("Committed in {} ms", duration);
+ } catch (CommitStateUnknownException commitStateUnknownException) {
+ cleanupOnAbort = false;
+ throw commitStateUnknownException;
+ }
}
private void abort(WriterCommitMessage[] messages) {
- Map<String, String> props = table.properties();
- Tasks.foreach(files(messages))
- .retry(PropertyUtil.propertyAsInt(props, COMMIT_NUM_RETRIES,
COMMIT_NUM_RETRIES_DEFAULT))
- .exponentialBackoff(
- PropertyUtil.propertyAsInt(props, COMMIT_MIN_RETRY_WAIT_MS,
COMMIT_MIN_RETRY_WAIT_MS_DEFAULT),
- PropertyUtil.propertyAsInt(props, COMMIT_MAX_RETRY_WAIT_MS,
COMMIT_MAX_RETRY_WAIT_MS_DEFAULT),
- PropertyUtil.propertyAsInt(props, COMMIT_TOTAL_RETRY_TIME_MS,
COMMIT_TOTAL_RETRY_TIME_MS_DEFAULT),
- 2.0 /* exponential */)
- .throwFailureWhenFinished()
- .run(file -> {
- table.io().deleteFile(file.path().toString());
- });
+ if (cleanupOnAbort) {
Review Comment:
clever way of working around the Spark interface on abort handling. I assume
a `SparkWrite` object is only used for a one-time write, right? If it can be
used for multiple writes/commits, we would need to reset this variable.
--
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]
