tomscut commented on a change in pull request #3887:
URL: https://github.com/apache/hadoop/pull/3887#discussion_r784777863
##########
File path:
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BPServiceActor.java
##########
@@ -1304,12 +1305,18 @@ long scheduleBlockReport(long delay, boolean
isRegistration) {
* Numerical overflow is possible here.
*/
void scheduleNextBlockReport() {
+ scheduleNextBlockReport(false);
+ }
+
+ void scheduleNextBlockReport(boolean forceFullBr) {
// If we have sent the first set of block reports, then wait a random
// time before we start the periodic block reports.
if (resetBlockReportTime) {
nextBlockReportTime.getAndSet(monotonicNow() +
ThreadLocalRandom.current().nextInt((int)
(blockReportIntervalMs)));
resetBlockReportTime = false;
+ } else if (forceFullBr) {
+ nextBlockReportTime.getAndSet(monotonicNow() + blockReportIntervalMs);
Review comment:
> I think that in most of cases, we want to know the exactly
nextBlockReportTime if we trigger block report by force. As for trigger full
block report in batches, I think it is a dangerous behavior because it may
cause some datanode heartbeat timeout.
I mean, if we trigger the FBR on some nodes in a short period of time, it
will affect the original randomness.
--
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]
