mvanhorn commented on code in PR #11102:
URL: https://github.com/apache/paimon/pull/11102#discussion_r4644862694


##########
paimon-core/src/main/java/org/apache/paimon/operation/FileStoreCommitImpl.java:
##########
@@ -1386,13 +1615,21 @@ private CommitResult tryCommitOnceWithoutLock(
         }
 
         // Check snapshot history even when the file-level conflict scan can 
be skipped.
-        List<BinaryRow> changedPartitions =
-                snapshotConflictChecker.check(
-                        newSnapshotId,
-                        commitKind,
-                        changes,
-                        rebasedReassignments,
-                        compactionBaselines);
+        List<BinaryRow> changedPartitions;
+        try {
+            changedPartitions =
+                    snapshotConflictChecker.check(
+                            newSnapshotId,
+                            commitKind,
+                            changes,
+                            rebasedReassignments,
+                            compactionBaselines);
+        } catch (BucketCompactionConflictException e) {

Review Comment:
   Agreed, an expired baseline needs its own path. In the next push, when a 
recovered COMPACT reaches checkBucketCompaction and its baseline plus one is 
below `snapshotManager.earliestSnapshotId()`, the checker will throw 
BucketCompactionConflictException so tryCommitCompact skips that recovery with 
a warning, while any other snapshot read failure still propagates and stays 
retryable. I'll add a test for your sequence (checkpoint 2 prepared on snapshot 
2, foreign APPEND as 3, its APPEND as 4, snapshots 1 to 3 expired) asserting 
both fresh restorations skip the compact phase and keep the APPEND.
   
   <!-- r:35439cf10f71.ce92b0942209 -->



-- 
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]

Reply via email to