bbeaudreault commented on code in PR #4536:
URL: https://github.com/apache/hbase/pull/4536#discussion_r1118825996
##########
hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/StoreScanner.java:
##########
@@ -768,7 +768,11 @@ public boolean next(List<Cell> outResult, ScannerContext
scannerContext) throws
// One last chance to break due to size limit. The INCLUDE* cases
above already check
// limit and continue. For the various filtered cases, we need to
check because block
// size limit may have been exceeded even if we don't add cells to
result list.
- if (scannerContext.checkSizeLimit(LimitScope.BETWEEN_CELLS)) {
+ // And when reaching the heartbeat cells, try to return from the loop.
+ if (
+ scannerContext.checkSizeLimit(LimitScope.BETWEEN_CELLS)
+ || kvsScanned % cellsPerHeartbeatCheck == 0
Review Comment:
I think the `kvsScanned % cellsPerHeartbeatCheck` was meant to avoid the
cost of checking time limit (which requires hitting the clock). You're re-using
it here as well to force compaction to check-in periodically based on
cells/bytes scanned. But that's actually the whole point behind the changes in
HBASE-27558, so that makes me think we should just apply the same change to
compactions by specificying a block size limit in Compator.
--
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]
