junrao commented on code in PR #15825:
URL: https://github.com/apache/kafka/pull/15825#discussion_r1586835182


##########
storage/src/main/java/org/apache/kafka/storage/internals/log/LogOffsetMetadata.java:
##########
@@ -28,6 +28,7 @@ public final class LogOffsetMetadata {
 
     //TODO KAFKA-14484 remove once UnifiedLog has been moved to the storage 
module
     private static final long UNIFIED_LOG_UNKNOWN_OFFSET = -1L;
+    public static final long REMOTE_LOG_UNKNOWN_OFFSET = -2L;

Review Comment:
   We probably don't need this. The existing UNIFIED_LOG_UNKNOWN_OFFSET should 
be enough.



##########
storage/src/main/java/org/apache/kafka/storage/internals/log/LogOffsetMetadata.java:
##########
@@ -51,6 +57,8 @@ public LogOffsetMetadata(long messageOffset,
 
     // check if this offset is already on an older segment compared with the 
given offset
     public boolean onOlderSegment(LogOffsetMetadata that) {
+        if (this.segmentBaseOffset == REMOTE_LOG_UNKNOWN_OFFSET || 
that.segmentBaseOffset == REMOTE_LOG_UNKNOWN_OFFSET)

Review Comment:
   We probably don't need this. If messageOffsetOnly() is true, we can just 
return false.



##########
storage/src/main/java/org/apache/kafka/storage/internals/log/LogOffsetMetadata.java:
##########
@@ -65,6 +73,8 @@ private boolean onSameSegment(LogOffsetMetadata that) {
     // compute the number of bytes between this offset to the given offset
     // if they are on the same segment and this offset precedes the given 
offset
     public int positionDiff(LogOffsetMetadata that) {
+        if (this.segmentBaseOffset == REMOTE_LOG_UNKNOWN_OFFSET || 
that.segmentBaseOffset == REMOTE_LOG_UNKNOWN_OFFSET)
+            return 1;

Review Comment:
   It's a bit hacky to do this here. I was thinking of doing this in 
DelayedFetch.
   
   ```
               if (endOffset.messageOffset != fetchOffset.messageOffset) {
                 if (endOffset.messageOnly() || fetchOffset.messageOnly()) {
                    accumulatedSize += 1
                 } else if (endOffset.onOlderSegment(fetchOffset)) {
                   // Case F, this can happen when the new fetch operation is 
on a truncated leader
   ...
               }
   ```



##########
core/src/main/scala/kafka/log/UnifiedLog.scala:
##########
@@ -1428,7 +1428,11 @@ class UnifiedLog(@volatile var logStartOffset: Long,
     */
   private def convertToOffsetMetadataOrThrow(offset: Long): LogOffsetMetadata 
= {
     checkLogStartOffset(offset)
-    localLog.convertToOffsetMetadataOrThrow(offset)
+    if (remoteLogEnabled() && offset < localLogStartOffset()) {
+      new LogOffsetMetadata(offset, 
LogOffsetMetadata.REMOTE_LOG_UNKNOWN_OFFSET)
+    } else {
+      localLog.convertToOffsetMetadataOrThrow(offset)

Review Comment:
   I was thinking that we could change 
localLog.convertToOffsetMetadataOrThrow() such that if read() throws an 
exception, it just returns a message-only LogOffsetMetadata.



-- 
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: jira-unsubscr...@kafka.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org

Reply via email to