yuqi1129 commented on code in PR #11892:
URL: https://github.com/apache/gravitino/pull/11892#discussion_r3528475421
##########
catalogs/catalog-lakehouse-generic/src/main/java/org/apache/gravitino/catalog/lakehouse/lance/LanceTableOperations.java:
##########
@@ -547,13 +552,45 @@ private Table repairTableMetadata(NameIdentifier ident,
Column[] columns, long d
}
}
+ /**
+ * Applies an idempotent update to the stored table, retrying when the
optimistic-lock CAS is lost
+ * to a concurrent update. The repair-on-load path runs on every {@code
loadTable}, so concurrent
+ * loads of the same table race on the version CAS; {@code store.update}
surfaces the lost race as
+ * an {@link IOException}. Because the updater is idempotent, the loser
re-reads the latest
+ * (already repaired) entity and retries instead of failing the whole load
with a fatal error.
+ */
+ private TableEntity updateTableWithCasRetry(
+ NameIdentifier ident, Function<TableEntity, TableEntity> updater) throws
IOException {
+ IOException lastConflict = null;
+ for (int attempt = 1; attempt <= REPAIR_UPDATE_MAX_ATTEMPTS; attempt++) {
+ try {
+ return store.update(ident, TableEntity.class, Entity.EntityType.TABLE,
updater);
+ } catch (IOException e) {
+ // Only retry when the update matched 0 rows (lost optimistic-lock
CAS). Other IO failures
+ // (DB outage, serialization errors, etc.) should fail fast.
+ String message = e.getMessage();
+ if (message == null || !message.startsWith("Failed to update the
entity:")) {
+ throw e;
+ }
Review Comment:
Fixed in 8ee3aa67fe: the literal is now a shared constant
`TableMetaService.UPDATE_ENTITY_CONFLICT_MESSAGE_PREFIX`, referenced by both
the thrower (`TableMetaService.updateTable`) and this retry, so the match has a
single source of truth instead of a duplicated string across the module
boundary.
A fuller structured signal (a dedicated `OptimisticLockFailureException`
emitted by the storage layer) would touch all 14 meta-services that share this
exact message, so I left that as a potential follow-up rather than expanding
this bugfix. Happy to file an issue for it if preferred.
--
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]
