[
https://issues.apache.org/jira/browse/HADOOP-19120?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17854394#comment-17854394
]
ASF GitHub Bot commented on HADOOP-19120:
-----------------------------------------
saxenapranav commented on PR #6633:
URL: https://github.com/apache/hadoop/pull/6633#issuecomment-2162706178
> Overall, I like the design, especially the way that it is possible to
switch back to the existing connections and in future move to the java11+ API.
>
> However, there is an absolute -1, KeepAliveCache contains code copied from
the JDK. You get to delete all code which is duplicate and review it very
carefully to make sure this is the case.
>
> I am not happy with us creating a long lived thread that never goes away.
Yes, it is done for the JVM, but that's a long-standing design decision of
theirs. Instead, make this one per abfs instance. When the filesystem is closed
this cache can/should be shut down. Or is there something I am missing here
-such as how it integrates with the JDK?
>
> As usual, I've complained about javadocs a lot. This is for the people
that come to maintain it in the future, yourself included -and for IDE popups.
Thank you @steveloughran for the review. Really appreciate your time in this.
I have now differentiated KeepAliveCache code from the JDK's implementation.
Now, each filesystem would have an instance of KeepAliveCache which maintains
connection pooling for that filesystem. The cache would have a timer that would
be running till the lifecycle of the filesystem. When filesystem is closed, the
cache gets closed, and the timer also gets closed.
I have taken the comments, requesting your kind review please. Thank you a
lot again for your time and insights.
Thanks!
> [ABFS]: ApacheHttpClient adaptation as network library
> ------------------------------------------------------
>
> Key: HADOOP-19120
> URL: https://issues.apache.org/jira/browse/HADOOP-19120
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/azure
> Affects Versions: 3.5.0
> Reporter: Pranav Saxena
> Assignee: Pranav Saxena
> Priority: Major
> Labels: pull-request-available
>
> Apache HttpClient is more feature-rich and flexible and gives application
> more granular control over networking parameter.
> ABFS currently relies on the JDK-net library. This library is managed by
> OpenJDK and has no performance problem. However, it limits the application's
> control over networking, and there are very few APIs and hooks exposed that
> the application can use to get metrics, choose which and when a connection
> should be reused. ApacheHttpClient will give important hooks to fetch
> important metrics and control networking parameters.
> A custom implementation of connection-pool is used. The implementation is
> adapted from the JDK8 connection pooling. Reasons for doing it:
> 1. PoolingHttpClientConnectionManager heuristic caches all the reusable
> connections it has created. JDK's implementation only caches limited number
> of connections. The limit is given by JVM system property
> "http.maxConnections". If there is no system-property, it defaults to 5.
> Connection-establishment latency increased with all the connections were
> cached. Hence, adapting the pooling heuristic of JDK netlib,
> 2. In PoolingHttpClientConnectionManager, it expects the application to
> provide `setMaxPerRoute` and `setMaxTotal`, which the implementation uses as
> the total number of connections it can create. For application using ABFS, it
> is not feasible to provide a value in the initialisation of the
> connectionManager. JDK's implementation has no cap on the number of
> connections it can have opened on a moment. Hence, adapting the pooling
> heuristic of JDK netlib,
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
