<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<HTML>
<HEAD>
<TITLE> [99s-extend] Cowboy HTTPS connection memory usage
</TITLE>
<LINK REL="Index" HREF="index.html" >
<LINK REL="made" HREF="mailto:extend%40lists.ninenines.eu?Subject=Re%3A%20%5B99s-extend%5D%20Cowboy%20HTTPS%20connection%20memory%20usage&In-Reply-To=%3CCAJ0zLRNud5vq9CnvZm8AgH1ZMPCzdkFe5tzdi5s13iP4HR6cuQ%40mail.gmail.com%3E">
<META NAME="robots" CONTENT="index,nofollow">
<style type="text/css">
pre {
white-space: pre-wrap; /* css-2.1, curent FF, Opera, Safari */
}
</style>
<META http-equiv="Content-Type" content="text/html; charset=us-ascii">
<LINK REL="Previous" HREF="000025.html">
<LINK REL="Next" HREF="000027.html">
</HEAD>
<BODY BGCOLOR="#ffffff">
<H1>[99s-extend] Cowboy HTTPS connection memory usage</H1>
<B>rambocoder</B>
<A HREF="mailto:extend%40lists.ninenines.eu?Subject=Re%3A%20%5B99s-extend%5D%20Cowboy%20HTTPS%20connection%20memory%20usage&In-Reply-To=%3CCAJ0zLRNud5vq9CnvZm8AgH1ZMPCzdkFe5tzdi5s13iP4HR6cuQ%40mail.gmail.com%3E"
TITLE="[99s-extend] Cowboy HTTPS connection memory usage">erlang at rambocoder.com
</A><BR>
<I>Fri Dec 21 20:25:10 CET 2012</I>
<P><UL>
<LI>Previous message: <A HREF="000025.html">[99s-extend] Cowboy HTTPS connection memory usage
</A></li>
<LI>Next message: <A HREF="000027.html">[99s-extend] [ANN] Ranch 0.6.0 Xmas Edition Released
</A></li>
<LI> <B>Messages sorted by:</B>
<a href="date.html#26">[ date ]</a>
<a href="thread.html#26">[ thread ]</a>
<a href="subject.html#26">[ subject ]</a>
<a href="author.html#26">[ author ]</a>
</LI>
</UL>
<HR>
<!--beginarticle-->
<PRE>Long story short, I solved the problem by adding {max_connections, 50000}
to cowboy:start_https because it default to 1024 at
<A HREF="https://github.com/extend/ranch/blob/master/src/ranch_listener_sup.erl#L30">https://github.com/extend/ranch/blob/master/src/ranch_listener_sup.erl#L30</A>
However, before I figured out that setting, I did run eprof and these are
the function calls it was spending most of it's time on
FUNCTION CALLS % TIME [uS /
CALLS]
-------- ----- --- ----
[----------]
dict:get_slot/2 174 1.73 1658 [
9.53]
dict:on_bucket/3 171 1.77 1701 [
9.95]
erlang:setelement/3 684 3.23 3098 [
4.53]
dict:store_bkt_val/3 600 5.24 5028 [
8.38]
Then I ran etop and it showed that ranch_acceptor:maybe_wait had the most
reductions were, so I looked at the code in that
<A HREF="https://github.com/extend/ranch/blob/master/src/ranch_acceptor.erl#L72">https://github.com/extend/ranch/blob/master/src/ranch_acceptor.erl#L72</A> and
realized that like a newb I did not set the maximum connections for the
listener :)
Problem solved. Looks like I won't need to put HAProxy in front of Cowboy
after all.
Thank you,
rambocoder
On Fri, Dec 21, 2012 at 11:51 AM, Loïc Hoguin <<A HREF="https://lists.ninenines.eu/listinfo/extend">essen at ninenines.eu</A>> wrote:
><i> Can you try enabling eprof to see where the VM spends its time?
</I>><i>
</I>><i>
</I>><i> On 12/21/2012 05:49 PM, rambocoder wrote:
</I>><i>
</I>>><i> In my preliminary testing, I used Jmeter this morning since it's an
</I>>><i> easy GUI load testing app and this is what I am seeing:
</I>>><i>
</I>>><i> With R15B03-01 [smp:4:4] [async-threads:4] [hipe] [kernel-poll:true],
</I>>><i> when I establish 1K concurrent connections via HTTPS, each connection
</I>>><i> takes up about 68K of memory.
</I>>><i>
</I>>><i> Unfortunately, after about 1050-1200 connections, on my test server the
</I>>><i> Erlang scheduler jumps to 100% CPU utilization on all 4 schedulers,
</I>>><i> while up to that point the scheduler's load was oscillating up and down.
</I>>><i> Using the Observer, there is only 1 ssl_connection_sup in the ssl
</I>>><i> application, having to deal with 1000+ gen_fsm workers, so that might be
</I>>><i> the bottleneck. Since the ulimit on my server is 50000 I don't think I
</I>>><i> am hitting any type of file handler's limit.
</I>>><i>
</I>>><i> Loïc and the group, am I missing some setting that is causing the
</I>>><i> scheduler to go to 100% CPU and the run que in observer to be 99?
</I>>><i>
</I>>><i> Sincerely,
</I>>><i>
</I>>><i> rambocoder
</I>>><i>
</I>>><i>
</I>>><i>
</I>>><i> On Fri, Dec 21, 2012 at 6:45 AM, Loïc Hoguin <<A HREF="https://lists.ninenines.eu/listinfo/extend">essen at ninenines.eu</A>
</I>>><i> <mailto:<A HREF="https://lists.ninenines.eu/listinfo/extend">essen at ninenines.eu</A>>> wrote:
</I>>><i>
</I>>><i> On 12/21/2012 04:34 AM, rambocoder wrote:
</I>>><i>
</I>>><i> Does anybody know either from benchmarks or real world data what
</I>>><i> is the
</I>>><i> average memory footprint of each concurrent HTTPS connection to
</I>>><i> cowboy?
</I>>><i>
</I>>><i>
</I>>><i> I don't have anything, sorry. I'm guessing it consumes a lot more
</I>>><i> than TCP though.
</I>>><i>
</I>>><i>
</I>>><i> SSL app in Erlang reuses SSL session-ids so I am not sure if the
</I>>><i> Apache
</I>>><i> Bench I test with reuses the session id or it does not.
</I>>><i>
</I>>><i>
</I>>><i> I wouldn't know, but I wouldn't trust Apache Bench doing the right
</I>>><i> thing. Any other benchmark tool usually works better in my experience.
</I>>><i>
</I>>><i>
</I>>><i> BTW, what makes an erlang api "documented" vs "undocumented". For
</I>>><i> example ssl:session_info/1 function here (
</I>>><i> <A HREF="https://github.com/erlang/otp/**__blob/maint/lib/ssl/src/ssl._**">https://github.com/erlang/otp/**__blob/maint/lib/ssl/src/ssl._**</A>
</I>>><i> _erl#L411<<A HREF="https://github.com/erlang/otp/__blob/maint/lib/ssl/src/ssl.__erl#L411">https://github.com/erlang/otp/__blob/maint/lib/ssl/src/ssl.__erl#L411</A>>
</I>>><i>
</I>>><i> <<A HREF="https://github.com/erlang/**otp/blob/maint/lib/ssl/src/**">https://github.com/erlang/**otp/blob/maint/lib/ssl/src/**</A>
</I>>><i> ssl.erl#L411<<A HREF="https://github.com/erlang/otp/blob/maint/lib/ssl/src/ssl.erl#L411">https://github.com/erlang/otp/blob/maint/lib/ssl/src/ssl.erl#L411</A>>
</I>>><i> >
</I>>><i> ) has
</I>>><i> a spec and a short doc, but session_info is not described
</I>>><i> <A HREF="http://www.erlang.org/doc/man/**__ssl.html<http://www.erlang.org/doc/man/__ssl.html">http://www.erlang.org/doc/man/**__ssl.html<http://www.erlang.org/doc/man/__ssl.html</A>>
</I>>><i>
</I>>><i> <<A HREF="http://www.erlang.org/doc/**man/ssl.html<http://www.erlang.org/doc/man/ssl.html">http://www.erlang.org/doc/**man/ssl.html<http://www.erlang.org/doc/man/ssl.html</A>>>
</I>>><i> .ssl:session_info/1 is
</I>>><i> a useful
</I>>><i> function to be able to track if the load generator is reusing
</I>>><i> the SSL
</I>>><i> session_id or it is generating new one, because that would make
</I>>><i> all the
</I>>><i> difference during measurement due to Erlang caching SSL sessions
</I>>><i> by default.
</I>>><i>
</I>>><i>
</I>>><i> The documentation is separate (they're not using edoc). It's perhaps
</I>>><i> not deemed useful enough for documenting it. I wouldn't worry about
</I>>><i> using it for measurements though.
</I>>><i>
</I>>><i> Try asking Ingela on the ML about it, perhaps they just forgot to
</I>>><i> document it.
</I>>><i>
</I>>><i> --
</I>>><i> Loďc Hoguin
</I>>><i> Erlang Cowboy
</I>>><i> Nine Nines
</I>>><i> <A HREF="http://ninenines.eu">http://ninenines.eu</A>
</I>>><i>
</I>>><i>
</I>>><i>
</I>>><i>
</I>>><i> ______________________________**_________________
</I>>><i> Extend mailing list
</I>>><i> <A HREF="https://lists.ninenines.eu/listinfo/extend">Extend at lists.ninenines.eu</A>
</I>>><i> <A HREF="http://lists.ninenines.eu:81/**listinfo/extend<http://lists.ninenines.eu:81/listinfo/extend">http://lists.ninenines.eu:81/**listinfo/extend<http://lists.ninenines.eu:81/listinfo/extend</A>>
</I>>><i>
</I>>><i>
</I>><i>
</I>><i> --
</I>><i> Loïc Hoguin
</I>><i>
</I>><i> Erlang Cowboy
</I>><i> Nine Nines
</I>><i> <A HREF="http://ninenines.eu">http://ninenines.eu</A>
</I>><i>
</I>><i>
</I>-------------- next part --------------
An HTML attachment was scrubbed...
URL: <<A HREF="http://lists.ninenines.eu/archives/extend/attachments/20121221/945f636e/attachment.html">http://lists.ninenines.eu/archives/extend/attachments/20121221/945f636e/attachment.html</A>>
</PRE>
<!--endarticle-->
<HR>
<P><UL>
<!--threads-->
<LI>Previous message: <A HREF="000025.html">[99s-extend] Cowboy HTTPS connection memory usage
</A></li>
<LI>Next message: <A HREF="000027.html">[99s-extend] [ANN] Ranch 0.6.0 Xmas Edition Released
</A></li>
<LI> <B>Messages sorted by:</B>
<a href="date.html#26">[ date ]</a>
<a href="thread.html#26">[ thread ]</a>
<a href="subject.html#26">[ subject ]</a>
<a href="author.html#26">[ author ]</a>
</LI>
</UL>
<hr>
<a href="https://lists.ninenines.eu/listinfo/extend">More information about the Extend
mailing list</a><br>
</body></html>