[
https://issues.apache.org/jira/browse/HDFS-17762?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17938580#comment-17938580
]
ASF GitHub Bot commented on HDFS-17762:
---------------------------------------
hfutatzhanghb commented on PR #7539:
URL: https://github.com/apache/hadoop/pull/7539#issuecomment-2754184949
> Hi @hfutatzhanghb In the invokeMethod method of RouterAsyncRpcClient, the
client's threadLocalContext is reset to the async handler that processes the
request. Why is there still this issue?
Hi, @KeeProMise . Thanks for reviewing. Please refer to method
RouterRpcClient#addClientInfoToCallerContext.
Suppose we never reset CallerContext when using ARR, the origContext in
method addClientInfoToCallerContext would never be null and will conduct the
below logic:
```java
if (origContext != null) {
for (String part : origContext.split(contextFieldSeparator)) {
String[] keyValue =
part.split(CallerContext.Builder.KEY_VALUE_SEPARATOR, 2);
if (keyValue.length == 2) {
builder.appendIfAbsent(keyValue[0], keyValue[1]);
} else if (keyValue.length == 1) {
builder.append(keyValue[0]);
}
}
}
```
> [ARR] Reset CallerContext information when async handler thread done
> --------------------------------------------------------------------
>
> Key: HDFS-17762
> URL: https://issues.apache.org/jira/browse/HDFS-17762
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: rbf
> Reporter: farmmamba
> Assignee: farmmamba
> Priority: Major
> Labels: pull-request-available
>
> CallerContext information may be wrong due to the async handler executor
--
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
