[
https://issues.apache.org/jira/browse/ROCKETMQ-243?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16083386#comment-16083386
]
ASF GitHub Bot commented on ROCKETMQ-243:
-----------------------------------------
Github user shroman commented on a diff in the pull request:
https://github.com/apache/incubator-rocketmq/pull/130#discussion_r126858457
--- Diff:
common/src/main/java/org/apache/rocketmq/common/protocol/route/BrokerData.java
---
@@ -37,15 +41,21 @@ public BrokerData(String cluster, String brokerName,
HashMap<Long, String> broke
this.brokerAddrs = brokerAddrs;
}
+ /**
+ * Selects a (preferably master) broker address from the registered
list.
+ * If the master's address cannot be found, a slave broker address is
selected in a random manner.
+ *
+ * @return Broker address.
+ */
public String selectBrokerAddr() {
- String value = this.brokerAddrs.get(MixAll.MASTER_ID);
- if (null == value) {
- for (Map.Entry<Long, String> entry :
this.brokerAddrs.entrySet()) {
- return entry.getValue();
- }
+ String addr = this.brokerAddrs.get(MixAll.MASTER_ID);
+
+ if (addr == null) {
+ List<Long> keys = new ArrayList<Long>(brokerAddrs.keySet());
+ return brokerAddrs.get(keys.get(random.nextInt(keys.size())));
--- End diff --
what exactly is the problem with the current approach?
> BrokerData#selectBrokerAddr() picks the 1st element in the list of addresses
> ----------------------------------------------------------------------------
>
> Key: ROCKETMQ-243
> URL: https://issues.apache.org/jira/browse/ROCKETMQ-243
> Project: Apache RocketMQ
> Issue Type: Improvement
> Components: rocketmq-commons
> Affects Versions: 4.1.0-incubating
> Reporter: Roman Shtykh
> Assignee: Roman Shtykh
> Priority: Minor
>
> I wonder if this can cause broker contention. Therefore I propose to pick the
> address at least in a random manner.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
