[ 
https://issues.apache.org/jira/browse/HDFS-17651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17913998#comment-17913998
 ] 

ASF GitHub Bot commented on HDFS-17651:
---------------------------------------

hfutatzhanghb commented on code in PR #7244:
URL: https://github.com/apache/hadoop/pull/7244#discussion_r1919727057


##########
hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/resources/hdfs-rbf-default.xml:
##########
@@ -49,13 +49,39 @@
   </property>
 
   <property>
-    <name>dfs.federation.router.rpc.async.enable</name>
+    <name>dfs.federation.router.async.rpc.enable</name>
     <value>false</value>
     <description>
       If true, router will process the RPC request asynchronously.
     </description>
   </property>
 
+  <property>
+    <name>dfs.federation.router.async.rpc.ns.handler.count</name>
+    <value>nsPlaceholder1:0,nsPlaceholder2:0</value>

Review Comment:
   fixed. Thanks.



##########
hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RBFConfigKeys.java:
##########
@@ -72,15 +72,23 @@ public class RBFConfigKeys extends 
CommonConfigurationKeysPublic {
   public static final String DFS_ROUTER_RPC_ENABLE =
       FEDERATION_ROUTER_PREFIX + "rpc.enable";
   public static final boolean DFS_ROUTER_RPC_ENABLE_DEFAULT = true;
-  public static final String DFS_ROUTER_RPC_ENABLE_ASYNC =
-      FEDERATION_ROUTER_PREFIX + "rpc.async.enable";
-  public static final boolean DFS_ROUTER_RPC_ENABLE_ASYNC_DEFAULT = false;
-  public static final String DFS_ROUTER_RPC_ASYNC_HANDLER_COUNT =
-      FEDERATION_ROUTER_PREFIX + "rpc.async.handler.count";
-  public static final int DFS_ROUTER_RPC_ASYNC_HANDLER_COUNT_DEFAULT = 2;
-  public static final String DFS_ROUTER_RPC_ASYNC_RESPONDER_COUNT =
-      FEDERATION_ROUTER_PREFIX + "rpc.async.responder.count";
-  public static final int DFS_ROUTER_RPC_ASYNC_RESPONDER_COUNT_DEFAULT = 10;
+  // HDFS Router Asynchronous RPC
+  public static final String DFS_ROUTER_ASYNC_RPC_ENABLE_KEY =
+      FEDERATION_ROUTER_PREFIX + "async.rpc.enable";
+  public static final boolean DFS_ROUTER_ASYNC_RPC_ENABLE_DEFAULT = false;
+  public static final String FEDERATION_ROUTER_ASYNC_RPC_PREFIX =
+          FEDERATION_ROUTER_PREFIX + "async.rpc.";
+  // Example: ns1:count1,ns2:count2,ns3:count3
+  public static final String DFS_ROUTER_ASYNC_RPC_NS_HANDLER_COUNT_KEY =
+          FEDERATION_ROUTER_ASYNC_RPC_PREFIX + "ns.handler.count";
+  public static final String DFS_ROUTER_ASYNC_RPC_NS_HANDLER_COUNT_DEFAULT =

Review Comment:
   Fixed.





> [ARR] Async handler executor isolation.
> ---------------------------------------
>
>                 Key: HDFS-17651
>                 URL: https://issues.apache.org/jira/browse/HDFS-17651
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>            Reporter: farmmamba
>            Assignee: farmmamba
>            Priority: Major
>              Labels: pull-request-available
>
> The main purpose of this PR is to isolate each nameservice by letting each 
> nameservice has its own async handler thread pool.
> Think below situation:
> When a downstream nameserivce ns1 has poor performance, the threads in async 
> handler thread pool will be occupied adding calls to the ns1's 
> Connection#calls. There will be no available threads to handle normal 
> nameservice's rpc request asynchronously. Therefore, it is better to isolate 
> the async handler thread pool of different nameservices.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org

Reply via email to