<div dir="ltr">The mapping is done by a hash of the remote IP+port. It sounds like maybe this machine is allocating ports in a way that does not map well to our hash. <div><br></div><div>Because the remote IP is the same it is really only the port that comes into effect. The algorithm is in org.xnio.nio.QueuedNioTcpServer#handleReady and in this case would simplify down to:</div><div><br></div><div>(((C1 * 23) + P) * 23 + C2) % 8</div><div><br></div><div>Where C1 is a hash of the remote IP, and C2 is a hash of the local IP+port combo. </div><div><div><div><br></div><div>Stuart</div></div></div></div><br><div class="gmail_quote"><div dir="ltr">On Thu, Jul 26, 2018 at 3:52 AM R. Matt Barnett <<a href="mailto:barnett@rice.edu">barnett@rice.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div text="#000000" bgcolor="#FFFFFF">
<p>I did. I set the concurrency level of ab to 128. I still see only
4 overlaps:</p>
<blockquote>
<p>$ java -jar undertow-test-0.1.0-jar-with-dependencies.jar &<br>
<br>
Server started on port 8080<br>
1<br>
2<br>
3<br>
4<br>
</p>
<p>$ netstat -t | grep apigateway_loadge | grep -c ESTABLISHED<br>
126</p>
</blockquote>
<p><br>
</p>
<p>What is the algorithm for mapping connections to IO threads? As
a new Undertow user I had assumed round robin, but it sounds like
this is not the case.</p>
<p><br>
</p>
<p>-- Matt<br>
</p>
<br>
<div class="m_-3278563139687851367moz-cite-prefix">On 7/25/2018 11:49 AM, Bill O'Neil
wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">Did you try setting the concurrency level much
higher than 8 like I suggested earlier? You are probably having
multiple connections assigned to the same IO threads.<input name="virtru-metadata" value="{"email-policy":{"state":"closed","expirationUnit":"days","disableCopyPaste":false,"disablePrint":false,"disableForwarding":false,"expires":false,"isManaged":false},"attachments":{},"compose-window":{"secure":false}}" type="hidden">
<div class="gmail_extra"><br>
<div class="gmail_quote">On Wed, Jul 25, 2018 at
12:26 PM, R. Matt Barnett <span dir="ltr"><<a href="mailto:barnett@rice.edu" target="_blank">barnett@rice.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Corrected
test to resolve test/set race.<br>
<br>
<br>
<a href="https://gist.github.com/rmbarnett-rice/1179c4ad1d3344bb247c8b8daed3e4fa" rel="noreferrer" target="_blank">https://gist.github.com/rmbarnett-rice/1179c4ad1d3344bb247c8b8daed3e4fa</a><br>
<br>
<br>
I've also discovered this morning that I *can* see 1-8
printed on Red <br>
Hat when I generate load using ab from Windows, but only
1-4 when <br>
running ab on Red Hat (both locally and from a remote
server). I'm <br>
wondering if perhaps there is some sort of connection
reuse shenanigans <br>
going on. My assumption of the use of the -c 8 parameter
was "make 8 <br>
sockets" but maybe not. I'll dig in and report back.<br>
<span class="m_-3278563139687851367HOEnZb"><font color="#888888"><br>
<br>
-- Matt<br>
</font></span>
<div class="m_-3278563139687851367HOEnZb">
<div class="m_-3278563139687851367h5"><br>
<br>
On 7/24/2018 6:56 PM, R. Matt Barnett wrote:<br>
> Hello,<br>
><br>
> I'm experiencing an Undertow performance issue I
fail to understand. I<br>
> am able to reproduce the issue with the code
linked bellow. The problem<br>
> is that on Red Hat (and not Windows) I'm unable
to concurrently process<br>
> more than 4 overlapping requests even with 8
configured IO Threads.<br>
> For example, if I run the following program (1
file, 55 lines):<br>
><br>
> <a href="https://gist.github.com/rmbarnett-rice/668db6b4e9f8f8da7093a3659b6ae2b5" rel="noreferrer" target="_blank">https://gist.github.com/rmbarnett-rice/668db6b4e9f8f8da7093a3659b6ae2b5</a><br>
><br>
> ... on Red Hat and then send requests to the
server using Apache<br>
> Benchmark...<br>
><br>
> > ab -n 1000 -c 8 localhost:8080/<br>
><br>
> I see the following output from the Undertow
process:<br>
><br>
> Server started on port 8080<br>
><br>
> 1<br>
> 2<br>
> 3<br>
> 4<br>
><br>
> I believe this demonstrates that only 4 requests
are ever processed in<br>
> parallel. I would expect 8. In fact, when I run
the same experiment on<br>
> Windows I see the expected output of<br>
><br>
> Server started on port 8080<br>
> 1<br>
> 2<br>
> 3<br>
> 4<br>
> 5<br>
> 6<br>
> 7<br>
> 8<br>
><br>
> Any thoughts as to what might explain this
behavior?<br>
><br>
> Best,<br>
><br>
> Matt<br>
><br>
> _______________________________________________<br>
> undertow-dev mailing list<br>
> <a href="mailto:undertow-dev@lists.jboss.org" target="_blank">undertow-dev@lists.jboss.org</a><br>
> <a href="https://lists.jboss.org/mailman/listinfo/undertow-dev" rel="noreferrer" target="_blank">https://lists.jboss.org/mailman/listinfo/undertow-dev</a><br>
<br>
_______________________________________________<br>
undertow-dev mailing list<br>
<a href="mailto:undertow-dev@lists.jboss.org" target="_blank">undertow-dev@lists.jboss.org</a><br>
<a href="https://lists.jboss.org/mailman/listinfo/undertow-dev" rel="noreferrer" target="_blank">https://lists.jboss.org/mailman/listinfo/undertow-dev</a></div>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</blockquote>
<br>
</div>
_______________________________________________<br>
undertow-dev mailing list<br>
<a href="mailto:undertow-dev@lists.jboss.org" target="_blank">undertow-dev@lists.jboss.org</a><br>
<a href="https://lists.jboss.org/mailman/listinfo/undertow-dev" rel="noreferrer" target="_blank">https://lists.jboss.org/mailman/listinfo/undertow-dev</a></blockquote></div>