puchengy commented on code in PR #12818:
URL: https://github.com/apache/iceberg/pull/12818#discussion_r2071888795
##########
core/src/main/java/org/apache/iceberg/rest/RESTTableOperations.java:
##########
@@ -155,13 +158,21 @@ public void commit(TableMetadata base, TableMetadata
metadata) {
// the error handler will throw necessary exceptions like
CommitFailedException and
// UnknownCommitStateException
// TODO: ensure that the HTTP client lib passes HTTP client errors to the
error handler
- LoadTableResponse response =
- client.post(path, request, LoadTableResponse.class, headers,
errorHandler);
+ try {
+ LoadTableResponse response =
+ client.post(path, request, LoadTableResponse.class, headers,
errorHandler);
+ // all future commits should be simple commits
+ this.updateType = UpdateType.SIMPLE;
- // all future commits should be simple commits
- this.updateType = UpdateType.SIMPLE;
-
- updateCurrentMetadata(response);
+ updateCurrentMetadata(response);
+ } catch (RESTException e) {
+ if (e.getCause() != null && e.getCause() instanceof IOException) {
+ // any IOException or unhandled Exception should be considered as
commit unknown
+ // so that caller can attempt to potentially reconcile
+ throw new CommitStateUnknownException(e);
+ }
+ throw e;
+ }
Review Comment:
@singhpk234 I took a second look and I found one difference between my
internal version (spark 3.2 + iceberg 1.3) and the latest version: the
`cleanupOnAbort` by default is `true` in the old version but it was set as
`false` in the new version. It seems to be introduced by
https://github.com/apache/iceberg/pull/10373 by @amogh-jahagirdar
--
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]
