[
https://issues.apache.org/jira/browse/CAMEL-10240?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15465611#comment-15465611
]
ASF GitHub Bot commented on CAMEL-10240:
----------------------------------------
GitHub user tkopczynski opened a pull request:
https://github.com/apache/camel/pull/1157
CAMEL-10240: Fix for thread pool sizes in CamelHttpClient
Fixes CAMEL-10240.
Things I did:
- Removed setting the `QueuedThreadPool` size to 16 in `CamelHttpClient`.
Now it relies on defaults in `QueuedThreadPool` class itself.
- Added a constructor in `CamelHttpClient` for setting
`HttpClientTransport`.
- If the `httpClientMaxThreads` parameter is set, it becomes the selectors
number for the `HttpClientTransport` instance. I'm not sure about this one but
I thought it might be good to enable the possibility to configure this instead
of having the dependency on `Runtime.getRuntime().availableProcessors()` for
all cases. But maybe it would be better to create a new parameter for this?
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/tkopczynski/camel camel-10240
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/camel/pull/1157.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #1157
----
commit 6f12d971fe55738ee6f1e96b33917cd230d8af99
Author: Tomasz Kopczynski <[email protected]>
Date: 2016-09-05T18:55:04Z
CAMEL-10240: Fix for thread pool sizes in CamelHttpClient
----
> CamelHttpClient default pool size does not work on system with large number
> of cpus
> -----------------------------------------------------------------------------------
>
> Key: CAMEL-10240
> URL: https://issues.apache.org/jira/browse/CAMEL-10240
> Project: Camel
> Issue Type: Improvement
> Components: camel-jetty
> Affects Versions: 2.17.2
> Reporter: Zoltan Farkas
>
> In CamelHttpClient a default threadpool is being created with a hardcoded
> size of 16:
> {code}
> @Override
> protected void doStart() throws Exception {
> if (!hasThreadPool()) {
> // if there is no thread pool then create a default thread pool
> using daemon threads
> QueuedThreadPool qtp = new QueuedThreadPool();
> // 16 max threads is the default in the http client
> qtp.setMaxThreads(16);
> qtp.setDaemon(true);
> // let the thread names indicate they are from the client
> qtp.setName("CamelJettyClient(" +
> ObjectHelper.getIdentityHashCode(this) + ")");
> setThreadPoolOrExecutor(qtp);
> }
> super.doStart();
> }
> {code}
> this configuration does not work because number of selectors is calculated
> relative to the number of CPUs in HttpClientTransportOverHTTP:
> {code}
> public HttpClientTransportOverHTTP()
> {
> this(Math.max(1, Runtime.getRuntime().availableProcessors() / 2));
> }
> {code}
> As a result by default camel simply hangs on these systems...
> It would be nice if CamelHttpClient would have a constructor that set the
> HttpClientTransport or at least the number of selectors.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
