[#59462] [ruby-trunk - Bug #9342][Open] [PATCH] SizedQueue#clear does not notify waiting threads in Ruby 1.9.3 — "jsc (Justin Collins)" <redmine@...>

9 messages 2014/01/02

[#59466] [ruby-trunk - Bug #9343][Open] [PATCH] SizedQueue#max= wakes up waiters properly — "normalperson (Eric Wong)" <normalperson@...>

11 messages 2014/01/02

[#59498] [ruby-trunk - Bug #9352][Open] [BUG] rb_sys_fail_str(connect(2) for [fe80::1%lo0]:3000) - errno == 0 — "kain (Claudio Poli)" <claudio@...>

10 messages 2014/01/03

[#59516] [ruby-trunk - Bug #9356][Open] TCPSocket.new does not seem to handle INTR — "charliesome (Charlie Somerville)" <charliesome@...>

48 messages 2014/01/03

[#59538] [ruby-trunk - Feature #9362][Assigned] Minimize cache misshit to gain optimal speed — "shyouhei (Shyouhei Urabe)" <shyouhei@...>

33 messages 2014/01/03
[#59582] Re: [ruby-trunk - Feature #9362][Assigned] Minimize cache misshit to gain optimal speed — SASADA Koichi <ko1@...> 2014/01/06

Intersting challenge.

[#59541] Re: [ruby-trunk - Feature #9362][Assigned] Minimize cache misshit to gain optimal speed — Eric Wong <normalperson@...> 2014/01/04

Hi, I noticed a trivial typo in array.c, and it fails building struct.c

[#59583] [ruby-trunk - Bug #9367][Open] REXML::XmlDecl doesn't use user specified quotes — "bearmini (Takashi Oguma)" <bear.mini@...>

12 messages 2014/01/06

[#59642] [ruby-trunk - Bug #9384][Open] Segfault in ruby 2.1.0p0 — "cbliard (Christophe Bliard)" <christophe.bliard@...>

11 messages 2014/01/08

[#59791] About unmarshallable DRb objects life-time — Rodrigo Rosenfeld Rosas <rr.rosas@...>

A while ago I created a proof-of-concept that I intended to use in my

16 messages 2014/01/15
[#59794] Re: About unmarshallable DRb objects life-time — Eric Hodel <drbrain@...7.net> 2014/01/15

On 15 Jan 2014, at 11:58, Rodrigo Rosenfeld Rosas <rr.rosas@gmail.com> wrote:

[#59808] Re: About unmarshallable DRb objects life-time — Rodrigo Rosenfeld Rosas <rr.rosas@...> 2014/01/16

Em 15-01-2014 19:42, Eric Hodel escreveu:

[#59810] Re: About unmarshallable DRb objects life-time — Eric Hodel <drbrain@...7.net> 2014/01/16

On 16 Jan 2014, at 02:15, Rodrigo Rosenfeld Rosas <rr.rosas@gmail.com> wrote:

[#59826] Re: About unmarshallable DRb objects life-time — Rodrigo Rosenfeld Rosas <rr.rosas@...> 2014/01/17

Em 16-01-2014 19:43, Eric Hodel escreveu:

[#59832] Re: About unmarshallable DRb objects life-time — Eric Hodel <drbrain@...7.net> 2014/01/17

On 17 Jan 2014, at 04:22, Rodrigo Rosenfeld Rosas <rr.rosas@gmail.com> wrote:

[ruby-core:60202] [ruby-trunk - Bug #9459] [Open] Net::HTTP initializes openssl library after TCP connection is established

From: josh.nw@...
Date: 2014-01-29 05:53:35 UTC
List: ruby-core #60202
Issue #9459 has been reported by Josh C.

----------------------------------------
Bug #9459: Net::HTTP initializes openssl library after TCP connection is established
https://bugs.ruby-lang.org/issues/9459

* Author: Josh C
* Status: Open
* Priority: Normal
* Assignee: Yui NARUSE
* Category: lib
* Target version: current: 2.2.0
* ruby -v: ruby 1.8.7 (2012-02-08 patchlevel 358) [universal-darwin12.0]
* Backport: 1.9.3: UNKNOWN, 2.0.0: UNKNOWN, 2.1: UNKNOWN
----------------------------------------
When making an HTTPS connection, the Net::HTTP#connect method makes the following calls:

1. TCPSocket.new
2. OpenSSL:SSL::SSLContext.new
3. OpenSSL:SSL:SSLSocket.connect

Here is a link to the [2.1.0](https://github.com/ruby/ruby/blob/v2_1_0/lib/net/http.rb#L877-L920) version, though the basic sequence is the same in trunk and as far back as 1.8.7, possibly earlier.

The problem is that between step 1 and 3, the server must keep the TCP socket open while the SSL client calls OpenSSL:SSL::SSLContext.new. The first time this code path is taken, step 2 has the side effect of initializing the OpenSSL library. This can take a non-trivial amount of time, and is made worse when several clients start at the same time.

I would suggest that the order of operations be switched to:

1. OpenSSL:SSL::SSLContext.new
2. TCPSocket.new
3. OpenSSL:SSL:SSLSocket.connect

I've attached a patch that shows this. It is based off of trunk.

Thank you

---Files--------------------------------
net_http_connect.patch (742 Bytes)


-- 
http://bugs.ruby-lang.org/

In This Thread

Prev Next