Apache9 commented on a change in pull request #3566:
URL: https://github.com/apache/hbase/pull/3566#discussion_r688842192
##########
File path:
hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/RSRpcServices.java
##########
@@ -308,18 +310,35 @@
*/
private static final long
DEFAULT_REGION_SERVER_RPC_MINIMUM_SCAN_TIME_LIMIT_DELTA = 10;
- /*
+ /**
* Whether to reject rows with size > threshold defined by
* {@link RSRpcServices#BATCH_ROWS_THRESHOLD_NAME}
*/
private static final String REJECT_BATCH_ROWS_OVER_THRESHOLD =
"hbase.rpc.rows.size.threshold.reject";
- /*
+ /**
* Default value of config {@link
RSRpcServices#REJECT_BATCH_ROWS_OVER_THRESHOLD}
*/
private static final boolean DEFAULT_REJECT_BATCH_ROWS_OVER_THRESHOLD =
false;
+ /**
+ * Determine the bootstrap nodes we want to return to the client connection
registry.
+ * <ul>
+ * <li>{@link #MASTER}: return masters as bootstrap nodes.</li>
Review comment:
There are no compatible issue between client and server...
The key difference between option 1 and 2, is that whether still support
using master as registry endpoint after the removal of MasterRegistry in 4.0.0.
If we still support this feature in the future, then I think a release note and
a notice email to dev list is enough. If we really think we should remove this
feature, then we'd better send a discussion email to dev list first. If there
are no big concerns, then we can do this. If no, I think we'd better go with
option 1, to let RpcConnectionRegistry still has the ability to use masters as
registry endpoint.
Is this clear enough? Let me point out the key problem again, it is about
whether people could still use masters as registry endpoint in 4.0.0, not about
now...
Thanks.
--
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]
