From normalperson at yhbt.net  Mon Mar  7 23:40:31 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Mon, 7 Mar 2011 20:40:31 -0800
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
Message-ID: <20110308044031.GA12532@dcvr.yhbt.net>

I started playing around with Bundler 1.0.10 today and noticed it's
quite different than previous versions (based on my limited experiences)
and the out-of-the-box experience is pretty good regarding (lack of)
ENV pollution.

Can any more experienced Bundler (and possibly Capistrano) users
comment on the below changes and see if they make sense?

Thanks in advance!

diff --git a/Sandbox b/Sandbox
index d101106..46dfb91 100644
--- a/Sandbox
+++ b/Sandbox
@@ -45,11 +45,20 @@ This is no longer be an issue as of bundler 0.9.17
 
 ref: http://mid.gmane.org/8FC34B23-5994-41CC-B5AF-7198EF06909E at tramchase.com
 
+=== BUNDLE_GEMFILE for Capistrano users
+
+You may need to set or reset the BUNDLE_GEMFILE environment variable in
+the before_exec hook:
+
+        before_exec do |server|
+          ENV["BUNDLE_GEMFILE"] = "/path/to/app/current/Gemfile"
+        end
+
 === Other ENV pollution issues
 
-You may need to set or reset BUNDLE_GEMFILE, GEM_HOME, GEM_PATH and PATH
-environment variables in the before_exec hook as illustrated by
-http://gist.github.com/534668
+If you're using an older Bundler version (0.9.x), you may need to set or
+reset GEM_HOME, GEM_PATH and PATH environment variables in the
+before_exec hook as illustrated by http://gist.github.com/534668
 
 == Isolate
 
-- 
Eric Wong

From justin.giancola at gmail.com  Tue Mar  8 09:22:32 2011
From: justin.giancola at gmail.com (Justin Giancola)
Date: Tue, 8 Mar 2011 09:22:32 -0500
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
In-Reply-To: <20110308044031.GA12532@dcvr.yhbt.net>
References: <20110308044031.GA12532@dcvr.yhbt.net>
Message-ID: <AANLkTinE+A9O-cApKB1F6-QOS_vsj_9sJXJzjKrwJJbP@mail.gmail.com>

Setting BUNDLE_GEMFILE in the before_exec hook is the only
Bundler-related Unicorn config I've needed for Bundler >= 1.0.3

For <= 1.0.2 not even this was necessary because the template
executables Bundler generates didn't fully resolve the BUNDLE_GEMFILE
path until 1.0.3


Justin

On Mon, Mar 7, 2011 at 11:40 PM, Eric Wong <normalperson at yhbt.net> wrote:
> I started playing around with Bundler 1.0.10 today and noticed it's
> quite different than previous versions (based on my limited experiences)
> and the out-of-the-box experience is pretty good regarding (lack of)
> ENV pollution.
>
> Can any more experienced Bundler (and possibly Capistrano) users
> comment on the below changes and see if they make sense?
>
> Thanks in advance!
>
> diff --git a/Sandbox b/Sandbox
> index d101106..46dfb91 100644
> --- a/Sandbox
> +++ b/Sandbox
> @@ -45,11 +45,20 @@ This is no longer be an issue as of bundler 0.9.17
>
> ?ref: http://mid.gmane.org/8FC34B23-5994-41CC-B5AF-7198EF06909E at tramchase.com
>
> +=== BUNDLE_GEMFILE for Capistrano users
> +
> +You may need to set or reset the BUNDLE_GEMFILE environment variable in
> +the before_exec hook:
> +
> + ? ? ? ?before_exec do |server|
> + ? ? ? ? ?ENV["BUNDLE_GEMFILE"] = "/path/to/app/current/Gemfile"
> + ? ? ? ?end
> +
> ?=== Other ENV pollution issues
>
> -You may need to set or reset BUNDLE_GEMFILE, GEM_HOME, GEM_PATH and PATH
> -environment variables in the before_exec hook as illustrated by
> -http://gist.github.com/534668
> +If you're using an older Bundler version (0.9.x), you may need to set or
> +reset GEM_HOME, GEM_PATH and PATH environment variables in the
> +before_exec hook as illustrated by http://gist.github.com/534668
>
> ?== Isolate
>
> --
> Eric Wong
> _______________________________________________
> Unicorn mailing list - mongrel-unicorn at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
> Do not quote signatures (like this one) or top post when replying
>

From lawrence.pit at gmail.com  Wed Mar  9 05:39:35 2011
From: lawrence.pit at gmail.com (Lawrence Pit)
Date: Wed, 09 Mar 2011 21:39:35 +1100
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
In-Reply-To: <20110308044031.GA12532@dcvr.yhbt.net>
References: <20110308044031.GA12532@dcvr.yhbt.net>
Message-ID: <4D7758E7.6020505@gmail.com>

Hi Eric,

I tested this out:

- in the before_exec and after_fork I log PATH GEM_HOME GEM_PATH 
BUNDLE_GEMFILE for inspection after deployment

- deploy latest release, make sure to shutdown unicorn, then start cleanly

- prune all releases, except the last one

- deploy latest release again (which only sets the BUNDLE_GEMFILE env var)

I've tried this a few times, and everytime I get something like this:

https://gist.github.com/331b0decab62fd58483c

If I revert back to setting the GEM_HOME, GEM_PATH and PATH in the 
before_exec it works fine.


PS. When deploying I use USR2 followed by QUIT


Cheers,
Lawrence

> I started playing around with Bundler 1.0.10 today and noticed it's
> quite different than previous versions (based on my limited experiences)
> and the out-of-the-box experience is pretty good regarding (lack of)
> ENV pollution.
>
> Can any more experienced Bundler (and possibly Capistrano) users
> comment on the below changes and see if they make sense?
>
> Thanks in advance!
>
> diff --git a/Sandbox b/Sandbox
> index d101106..46dfb91 100644
> --- a/Sandbox
> +++ b/Sandbox
> @@ -45,11 +45,20 @@ This is no longer be an issue as of bundler 0.9.17
>
>   ref: http://mid.gmane.org/8FC34B23-5994-41CC-B5AF-7198EF06909E at tramchase.com
>
> +=== BUNDLE_GEMFILE for Capistrano users
> +
> +You may need to set or reset the BUNDLE_GEMFILE environment variable in
> +the before_exec hook:
> +
> +        before_exec do |server|
> +          ENV["BUNDLE_GEMFILE"] = "/path/to/app/current/Gemfile"
> +        end
> +
>   === Other ENV pollution issues
>
> -You may need to set or reset BUNDLE_GEMFILE, GEM_HOME, GEM_PATH and PATH
> -environment variables in the before_exec hook as illustrated by
> -http://gist.github.com/534668
> +If you're using an older Bundler version (0.9.x), you may need to set or
> +reset GEM_HOME, GEM_PATH and PATH environment variables in the
> +before_exec hook as illustrated by http://gist.github.com/534668
>
>   == Isolate
>


From normalperson at yhbt.net  Wed Mar  9 19:49:08 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Thu, 10 Mar 2011 00:49:08 +0000
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
In-Reply-To: <AANLkTinE+A9O-cApKB1F6-QOS_vsj_9sJXJzjKrwJJbP@mail.gmail.com>
References: <20110308044031.GA12532@dcvr.yhbt.net>
	<AANLkTinE+A9O-cApKB1F6-QOS_vsj_9sJXJzjKrwJJbP@mail.gmail.com>
Message-ID: <20110310004908.GA1088@dcvr.yhbt.net>

Justin Giancola <justin.giancola at gmail.com> wrote:
> Setting BUNDLE_GEMFILE in the before_exec hook is the only
> Bundler-related Unicorn config I've needed for Bundler >= 1.0.3

Thanks, that seems to match my observations, too.

From normalperson at yhbt.net  Wed Mar  9 19:52:27 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Thu, 10 Mar 2011 00:52:27 +0000
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
In-Reply-To: <4D7758E7.6020505@gmail.com>
References: <20110308044031.GA12532@dcvr.yhbt.net> <4D7758E7.6020505@gmail.com>
Message-ID: <20110310005227.GA1002@dcvr.yhbt.net>

Lawrence Pit <lawrence.pit at gmail.com> wrote:
> Hi Eric,
>
> I tested this out:

Thanks for taking your time to test.

> - in the before_exec and after_fork I log PATH GEM_HOME GEM_PATH  
> BUNDLE_GEMFILE for inspection after deployment
>
> - deploy latest release, make sure to shutdown unicorn, then start cleanly
>
> - prune all releases, except the last one
>
> - deploy latest release again (which only sets the BUNDLE_GEMFILE env var)
>
> I've tried this a few times, and everytime I get something like this:
>
> https://gist.github.com/331b0decab62fd58483c

Yikes, it looks like the UNICORN_FD env is getting clobbered somehow.
Not sure what could be causing that.

> If I revert back to setting the GEM_HOME, GEM_PATH and PATH in the  
> before_exec it works fine.

Which Bundler version is this with?

> PS. When deploying I use USR2 followed by QUIT

From lawrence.pit at gmail.com  Wed Mar  9 22:30:14 2011
From: lawrence.pit at gmail.com (Lawrence Pit)
Date: Thu, 10 Mar 2011 14:30:14 +1100
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
In-Reply-To: <20110310005227.GA1002@dcvr.yhbt.net>
References: <20110308044031.GA12532@dcvr.yhbt.net> <4D7758E7.6020505@gmail.com>
	<20110310005227.GA1002@dcvr.yhbt.net>
Message-ID: <4D7845C6.2070301@gmail.com>


 >  Which Bundler version is this with?

Bundler 1.0.10


Cheers,
Lawrence

> I've tried this a few times, and everytime I get something like this:
>> https://gist.github.com/331b0decab62fd58483c
> Yikes, it looks like the UNICORN_FD env is getting clobbered somehow.
> Not sure what could be causing that.
>
>> If I revert back to setting the GEM_HOME, GEM_PATH and PATH in the
>> before_exec it works fine.

From julien.palmas at gmail.com  Thu Mar 10 06:21:14 2011
From: julien.palmas at gmail.com (Julien Palmas)
Date: Thu, 10 Mar 2011 12:21:14 +0100
Subject: Hello and Rack Handler question ;-)
Message-ID: <C5FE7048-F606-4533-96AE-00825DDD6A29@gmail.com>

Hi list,

This is my first post here, so Hello World! ;-)

By default, `rails server` starts webrick on a new rails 3 project.
If you include Thin in the gemfile, you can run `rails s thin`to use thin instead.
Same works with mongrel.

Unfortunately, this does not work with unicorn, since it seems that Rack does not provide a handler for Unicorn by default (as it does for Thin and Mongrel).

I know I can run `unicorn` in my app root to achieve the same, but it'd be nice to stick with rails commands and have `rails s unicorn`

Can you tell me if such a handler already exists ?

++
Ju


From list at mrtech.ru  Thu Mar 10 12:16:56 2011
From: list at mrtech.ru (Troex Nevelin)
Date: Thu, 10 Mar 2011 20:16:56 +0300
Subject: Memcache reset on rails 2.3 with preload
Message-ID: <4D790788.8010305@mrtech.ru>

What is a right code for resetting memcache connection in after_fork for 
Rails 2.3 with "preload_app true" on REE?

The only configurations about memcache in my app are:

config/initializers/session_store.rb:
ActionController::Base.session_store = :mem_cache_store

config/environment.rb:
config.cache_store = :mem_cache_store


I tried:
Rails.cache.reset

But it fails to start, looks like this code is for Rails3 only

From cliftonk at gmail.com  Thu Mar 10 15:48:22 2011
From: cliftonk at gmail.com (Clifton King)
Date: Thu, 10 Mar 2011 14:48:22 -0600
Subject: Memcache reset on rails 2.3 with preload
In-Reply-To: <4D790788.8010305@mrtech.ru>
References: <4D790788.8010305@mrtech.ru>
Message-ID: <AANLkTik_90ERYfZgQp0PSWZ6DQE54jQvtWjqZbL+2-ZS@mail.gmail.com>

I opt to use this gem: https://github.com/fauna/memcached

I use the following settings in my environment:

  $memcached         = Memcached::Rails.new(cache_server,
                                            :namespace       => "os-",
                                            :tcp_nodelay     => true,
  # disable nagle algorithm
                                            :no_block        => true)
  # dont block on writes

ActionController::Base.cache_store                = :mem_cache_store, $memcached
  $session_memcached = Memcached::Rails.new(cache_server,
                                            :namespace       => "os-",
                                            :tcp_nodelay     => true,
  # disable nagle algorithm
                                            :no_block        => true)
  ActionController::Base.cache_store                =
:mem_cache_store, $memcached
  ActionController::Base.session_store              = :mem_cache_store
  ActionController::Base.session_options[:cache]    = $session_memcached


in after_fork i have

$memcached.reset
$session_memcached.reset


Clifton

On Thu, Mar 10, 2011 at 11:16 AM, Troex Nevelin <list at mrtech.ru> wrote:
> What is a right code for resetting memcache connection in after_fork for
> Rails 2.3 with "preload_app true" on REE?
>
> The only configurations about memcache in my app are:
>
> config/initializers/session_store.rb:
> ActionController::Base.session_store = :mem_cache_store
>
> config/environment.rb:
> config.cache_store = :mem_cache_store
>
>
> I tried:
> Rails.cache.reset
>
> But it fails to start, looks like this code is for Rails3 only
> _______________________________________________
> Unicorn mailing list - mongrel-unicorn at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
> Do not quote signatures (like this one) or top post when replying
>

From normalperson at yhbt.net  Thu Mar 10 16:08:01 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Thu, 10 Mar 2011 13:08:01 -0800
Subject: Hello and Rack Handler question ;-)
In-Reply-To: <C5FE7048-F606-4533-96AE-00825DDD6A29@gmail.com>
References: <C5FE7048-F606-4533-96AE-00825DDD6A29@gmail.com>
Message-ID: <20110310210801.GA8221@dcvr.yhbt.net>

Julien Palmas <julien.palmas at gmail.com> wrote:
> Hi list,
> 
> This is my first post here, so Hello World! ;-)

Hello Julien :>

> By default, `rails server` starts webrick on a new rails 3 project.
> If you include Thin in the gemfile, you can run `rails s thin`to use
> thin instead.
> Same works with mongrel.
> 
> Unfortunately, this does not work with unicorn, since it seems that
> Rack does not provide a handler for Unicorn by default (as it does for
> Thin and Mongrel).
> 
> I know I can run `unicorn` in my app root to achieve the same, but
> it'd be nice to stick with rails commands and have `rails s unicorn`
> 
> Can you tell me if such a handler already exists ?

No it does not.  Unicorn provides configuration options and signal
handler interfaces that Thin and Mongrel do not.  Trying to tie them to
the same interface would just cripple Unicorn users attempting to use
the generic interface and make it more confusing and difficult to
support for everyone (Rack/Rails/Unicorn developers).

There was a similar discussion here back in 2009:
http://thread.gmane.org/gmane.comp.lang.ruby.unicorn.general/130/focus=130

-- 
Eric Wong

From normalperson at yhbt.net  Thu Mar 10 16:23:10 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Thu, 10 Mar 2011 13:23:10 -0800
Subject: Memcache reset on rails 2.3 with preload
In-Reply-To: <4D790788.8010305@mrtech.ru>
References: <4D790788.8010305@mrtech.ru>
Message-ID: <20110310212310.GA7801@dcvr.yhbt.net>

Troex Nevelin <list at mrtech.ru> wrote:
> What is a right code for resetting memcache connection in after_fork for  
> Rails 2.3 with "preload_app true" on REE?
>
> The only configurations about memcache in my app are:
>
> config/initializers/session_store.rb:
> ActionController::Base.session_store = :mem_cache_store
>
> config/environment.rb:
> config.cache_store = :mem_cache_store

So that uses the memcache-client gem?  I seem to remember that (and
dalli) only connects when it's needed, but I suppose some apps use
memcached at load time.

The following should work, not the most elegant:

  before_fork do |server,worker|
    ObjectSpace.each_object(MemCache) { |mc| mc.reset }
  end

An after_fork would probably send unnecessary messages to the memcached
servers.

> I tried:
> Rails.cache.reset
>
> But it fails to start, looks like this code is for Rails3 only

Hopefully somebody else knows a more elegant way to handle this.

-- 
Eric Wong

From normalperson at yhbt.net  Thu Mar 10 16:29:03 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Thu, 10 Mar 2011 13:29:03 -0800
Subject: [RFC/PATCH] Bundler/Sandbox documentation updates
In-Reply-To: <4D7845C6.2070301@gmail.com>
References: <20110308044031.GA12532@dcvr.yhbt.net> <4D7758E7.6020505@gmail.com>
	<20110310005227.GA1002@dcvr.yhbt.net> <4D7845C6.2070301@gmail.com>
Message-ID: <20110310212903.GB7801@dcvr.yhbt.net>

Lawrence Pit <lawrence.pit at gmail.com> wrote:
> Eric Wong <normalperson at yhbt.net> wrote:
>> I've tried this a few times, and everytime I get something like this:
>>> https://gist.github.com/331b0decab62fd58483c
>> Yikes, it looks like the UNICORN_FD env is getting clobbered somehow.
>> Not sure what could be causing that.
>>
>>> If I revert back to setting the GEM_HOME, GEM_PATH and PATH in the
>>> before_exec it works fine.

> >  Which Bundler version is this with?
>
> Bundler 1.0.10

Odd, I actually used this in my Rainbows! config file to switch an
Isolate deploy to Bundler (1.0.10) yesterday.   Everything for Rainbows!
applies to Unicorn, too).

  # switching from Isolate to Bundler:
  if ENV["GEM_HOME"] =~ %r{/isolate/}
    ENV.delete "GEM_HOME"
    ENV.delete "GEM_PATH"
    # don't need anything else in $PATH for a web server
    ENV["PATH"] = "/home/ew/ruby-1.9.2/bin"

    # START_CTX is considered a stable interface in Unicorn
    start_ctx = Unicorn::HttpServer::START_CTX
    start_ctx[0] = "bundle"

    # it's even possible to use USR2 to switch between Unicorn and
    # Rainbows! without any downtime :)
    start_ctx[:argv] = %w(exec rainbows).concat(start_ctx[:argv])
  end

-- 
Eric Wong

From lawrence.pit at gmail.com  Thu Mar 10 17:01:24 2011
From: lawrence.pit at gmail.com (Lawrence Pit)
Date: Fri, 11 Mar 2011 09:01:24 +1100
Subject: Memcache reset on rails 2.3 with preload
In-Reply-To: <4D790788.8010305@mrtech.ru>
References: <4D790788.8010305@mrtech.ru>
Message-ID: <4D794A34.4090900@gmail.com>


I use:

if Rails.cache.is_a?(ActiveSupport::Cache::MemCacheStore)
   Rails.cache.instance_variable_get(:@data).reset
end



Cheers,
Lawrence

> What is a right code for resetting memcache connection in after_fork 
> for Rails 2.3 with "preload_app true" on REE?
>
> The only configurations about memcache in my app are:
>
> config/initializers/session_store.rb:
> ActionController::Base.session_store = :mem_cache_store
>
> config/environment.rb:
> config.cache_store = :mem_cache_store
>
>
> I tried:
> Rails.cache.reset
>
> But it fails to start, looks like this code is for Rails3 only
> _______________________________________________
> Unicorn mailing list - mongrel-unicorn at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
> Do not quote signatures (like this one) or top post when replying
>


From cliftonk at gmail.com  Thu Mar 10 17:14:27 2011
From: cliftonk at gmail.com (Clifton King)
Date: Thu, 10 Mar 2011 16:14:27 -0600
Subject: Memcache reset on rails 2.3 with preload
In-Reply-To: <20110310212310.GA7801@dcvr.yhbt.net>
References: <4D790788.8010305@mrtech.ru> <20110310212310.GA7801@dcvr.yhbt.net>
Message-ID: <AANLkTikZ=g4yq=J2F5G9_yWXiKYctdtjwbedx8vwi66u@mail.gmail.com>

Eric,

No, I'm using the "memcached" gem which has a binding for libmemcached
(and is much faster than the memcache-client gem).

The memcached gem has a compatibility wrapper (Memcached::Rails) which
is made to be backwards compatible with memcache-client.

Clifton

On Thu, Mar 10, 2011 at 3:23 PM, Eric Wong <normalperson at yhbt.net> wrote:
> Troex Nevelin <list at mrtech.ru> wrote:
>> What is a right code for resetting memcache connection in after_fork for
>> Rails 2.3 with "preload_app true" on REE?
>>
>> The only configurations about memcache in my app are:
>>
>> config/initializers/session_store.rb:
>> ActionController::Base.session_store = :mem_cache_store
>>
>> config/environment.rb:
>> config.cache_store = :mem_cache_store
>
> So that uses the memcache-client gem? ?I seem to remember that (and
> dalli) only connects when it's needed, but I suppose some apps use
> memcached at load time.
>
> The following should work, not the most elegant:
>
> ?before_fork do |server,worker|
> ? ?ObjectSpace.each_object(MemCache) { |mc| mc.reset }
> ?end
>
> An after_fork would probably send unnecessary messages to the memcached
> servers.
>
>> I tried:
>> Rails.cache.reset
>>
>> But it fails to start, looks like this code is for Rails3 only
>
> Hopefully somebody else knows a more elegant way to handle this.
>
> --
> Eric Wong
> _______________________________________________
> Unicorn mailing list - mongrel-unicorn at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
> Do not quote signatures (like this one) or top post when replying
>

From empresamaravilha at hotmail.com  Sat Mar 12 07:13:16 2011
From: empresamaravilha at hotmail.com (Vascon)
Date: Sat, 12 Mar 2011 09:13:16 -0300
Subject: VOCÊ GOSTA DE INTERNET?
Message-ID: <mailman.0.1299932012.8658.mongrel-unicorn@rubyforge.org>

VOC? GOSTA DE INTERNET?
J? ganhou algum dinheiro com ela?
Trabalhe em suas horas vagas, sem atrapalhar sua
atual ocupa??o. SUPER HONESTO! MUITO LUCRATIVO!
SIMPLES! N?O PRECISA INSTALAR NADA!
Sem Consumir Produtos ou Fazer Reuni?es.
N?O TEM MENSALIDADES!
 Ficou curioso? Acesse e descubra...
www.macleusrendaextra.com.br/3905

From normalperson at yhbt.net  Tue Mar 15 08:32:41 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Tue, 15 Mar 2011 12:32:41 +0000
Subject: [ANN] unicorn 3.5.0 - very minor improvements
Message-ID: <20110315123241.GA11932@dcvr.yhbt.net>

Changes:

A small set of small changes but it's been more than a month
since our last release.  There are minor memory usage and
efficiently improvements (for graceful shutdowns).  MRI 1.8.7
users on *BSD should be sure they're using the latest patchlevel
(or upgrade to 1.9.x) because we no longer workaround their
broken stdio (that's MRI's job :)

* http://unicorn.bogomips.org/
* mongrel-unicorn at rubyforge.org
* git://bogomips.org/unicorn.git

-- 
Eric Wong

From normalperson at yhbt.net  Tue Mar 15 19:48:36 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Tue, 15 Mar 2011 16:48:36 -0700
Subject: Anybody running any Unicorn 3.x versions?
Message-ID: <20110315234836.GA22839@dcvr.yhbt.net>

Just wondering, are most people still using 1.1.x or have you started
using 3.x, yet?  If you're using Ruby 1.9.2, I expect 3.x to be better
(unless you have other things constantly wiping away your method cache
..)[1]

I've only heard a little feedback regarding breakage[2],
so either nobody's running it or it just works :)

[1] backport r28813 and r29673 in Ruby trunk to prevent that
[2] portability issues due to Kgio, but those are fixed

-- 
Eric Wong

From ghazel at gmail.com  Tue Mar 15 21:27:01 2011
From: ghazel at gmail.com (Greg Hazel)
Date: Tue, 15 Mar 2011 18:27:01 -0700
Subject: Anybody running any Unicorn 3.x versions?
In-Reply-To: <20110315234836.GA22839@dcvr.yhbt.net>
References: <20110315234836.GA22839@dcvr.yhbt.net>
Message-ID: <-530372658282867050@unknownmsgid>

Running Unicorn 3 here, the one you just released, on REE 1.8.7 in
production with no problems.

-Greg

On Mar 15, 2011, at 6:23 PM, Eric Wong <normalperson at yhbt.net> wrote:

> Just wondering, are most people still using 1.1.x or have you started
> using 3.x, yet?  If you're using Ruby 1.9.2, I expect 3.x to be better
> (unless you have other things constantly wiping away your method cache
> ..)[1]
>
> I've only heard a little feedback regarding breakage[2],
> so either nobody's running it or it just works :)
>
> [1] backport r28813 and r29673 in Ruby trunk to prevent that
> [2] portability issues due to Kgio, but those are fixed
>
> --
> Eric Wong
> _______________________________________________
> Unicorn mailing list - mongrel-unicorn at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
> Do not quote signatures (like this one) or top post when replying

From cliftonk at gmail.com  Tue Mar 15 21:59:24 2011
From: cliftonk at gmail.com (Clifton King)
Date: Tue, 15 Mar 2011 20:59:24 -0500
Subject: Anybody running any Unicorn 3.x versions?
In-Reply-To: <-530372658282867050@unknownmsgid>
References: <20110315234836.GA22839@dcvr.yhbt.net>
	<-530372658282867050@unknownmsgid>
Message-ID: <AANLkTikTPqQUQPw_g3g0Hg+L30VCpshs7J-Bn7RviPRV@mail.gmail.com>

Running unicorn 3.4 on 1.8.7.

On Tue, Mar 15, 2011 at 8:27 PM, Greg Hazel <ghazel at gmail.com> wrote:
> Running Unicorn 3 here, the one you just released, on REE 1.8.7 in
> production with no problems.
>
> -Greg
>
> On Mar 15, 2011, at 6:23 PM, Eric Wong <normalperson at yhbt.net> wrote:
>
>> Just wondering, are most people still using 1.1.x or have you started
>> using 3.x, yet? ?If you're using Ruby 1.9.2, I expect 3.x to be better
>> (unless you have other things constantly wiping away your method cache
>> ..)[1]
>>
>> I've only heard a little feedback regarding breakage[2],
>> so either nobody's running it or it just works :)
>>
>> [1] backport r28813 and r29673 in Ruby trunk to prevent that
>> [2] portability issues due to Kgio, but those are fixed
>>
>> --
>> Eric Wong
>> _______________________________________________
>> Unicorn mailing list - mongrel-unicorn at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
>> Do not quote signatures (like this one) or top post when replying
> _______________________________________________
> Unicorn mailing list - mongrel-unicorn at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-unicorn
> Do not quote signatures (like this one) or top post when replying
>

From normalperson at yhbt.net  Fri Mar 18 14:09:02 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Fri, 18 Mar 2011 11:09:02 -0700
Subject: [ANN] raindrops updates (mainly for Linux users)
Message-ID: <20110318180902.GA6076@dcvr.yhbt.net>

I figured I should cross post here.  raindrops (listen queue stats
package for Unicorn/Rainbows!) has gotten some largish updates over the
past few weeks:

raindrops 0.5.0 release announcement:
  http://mid.gmane.org/20110317033547.GA30653 at dcvr.yhbt.net

demo site changes (will probably be in 0.6.0 soon):
  http://mid.gmane.org/20110318094637.GA3526 at dcvr.yhbt.net

Feedback would be greatly appreciated (either here or
raindrops at librelist.com).  There are no backwards-incompatible changes
for existing users as far as I'm aware of.

Thanks for reading!

-- 
Eric Wong

From chenggn at gmail.com  Mon Mar 21 03:37:50 2011
From: chenggn at gmail.com (Guang-Nan Cheng)
Date: Mon, 21 Mar 2011 15:37:50 +0800
Subject: Does unicorn support mount the webapp with a prefix like Thin does?
In-Reply-To: <AANLkTinDS86owzdni8bgxN4RE=B=ocXAsqzcdH1u93U2@mail.gmail.com>
References: <AANLkTinDS86owzdni8bgxN4RE=B=ocXAsqzcdH1u93U2@mail.gmail.com>
Message-ID: <AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>

For example, I have a webapp that handles /welcome.

Does unicorn support to mount the entire webapp at /prefix so request
to /prefix/welcome are?handled?

From edhoprima at gmail.com  Mon Mar 21 04:34:35 2011
From: edhoprima at gmail.com (Edho P Arief)
Date: Mon, 21 Mar 2011 15:34:35 +0700
Subject: Does unicorn support mount the webapp with a prefix like Thin
	does?
In-Reply-To: <AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>
References: <AANLkTinDS86owzdni8bgxN4RE=B=ocXAsqzcdH1u93U2@mail.gmail.com>
	<AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>
Message-ID: <AANLkTimnKwZ=nbv-hcC=9k02w2m1HrcB8M1r-t2oYrOr@mail.gmail.com>

On Mon, Mar 21, 2011 at 2:37 PM, Guang-Nan Cheng <chenggn at gmail.com> wrote:
> For example, I have a webapp that handles /welcome.
>
> Does unicorn support to mount the entire webapp at /prefix so request
> to /prefix/welcome are?handled?

yes, it does. I don't remember exactly how though as my main server is
currently offline (it has the config and all). Probably using --path
or something.

From normalperson at yhbt.net  Mon Mar 21 12:14:10 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Mon, 21 Mar 2011 09:14:10 -0700
Subject: Does unicorn support mount the webapp with a prefix like Thin
	does?
In-Reply-To: <AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>
References: <AANLkTinDS86owzdni8bgxN4RE=B=ocXAsqzcdH1u93U2@mail.gmail.com>
	<AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>
Message-ID: <20110321161410.GA26947@dcvr.yhbt.net>

Guang-Nan Cheng <chenggn at gmail.com> wrote:
> For example, I have a webapp that handles /welcome.
> 
> Does unicorn support to mount the entire webapp at /prefix so request
> to /prefix/welcome are?handled?

Yes, all Rack servers that support rackup (config.ru) files can do it:

  map "/prefix" do
    use Rack::Chunked
    run WelcomeApp.new
  end
  # ...
  map "/lobster" do
    use Rack::Lint
    run Rack::Lobster.new
  end
  # You can do virtual hosts, too:
  map "http://example.com/" do
    run ExampleApp.new
  end

See the Rack::Builder and Rack::URLMap RDoc for more info.

-- 
Eric Wong

From david at zendesk.com  Tue Mar 22 15:19:36 2011
From: david at zendesk.com (David Nghiem)
Date: Tue, 22 Mar 2011 12:19:36 -0700
Subject: unicorn 1.1.x never-ending listen loop IOError exceptions
Message-ID: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>

We run unicorn 1.1.5 behind nginx under Ruby Enterprise Edition
2011.02.  We're using UNIX sockets instead of TCP sockets.

Once in a while, one worker will throw exceptions in a loop, taking up
lots of CPU and log space.  Killing the worker removes the problem.
Other workers are not affected.

Here's the exception:

E, [2011-01-10T10:13:13.518795 #15209] ERROR -- : Unhandled listen
loop exception #<IOError: closed stream>.
E, [2011-01-10T10:13:13.518881 #15209] ERROR -- :
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:732:in
`select'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:732:in
`worker_loop'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/newrelic_rpm-2.13.1/lib/new_relic/control/../agent/instrumentation/unicorn_instrumentation.rb:7:in
`call'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/newrelic_rpm-2.13.1/lib/new_relic/control/../agent/instrumentation/unicorn_instrumentation.rb:7:in
`worker_loop'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:603:in
`spawn_missing_workers'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:600:in
`fork'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:600:in
`spawn_missing_workers'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:596:in
`each'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:596:in
`spawn_missing_workers'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:610:in
`maintain_worker_count'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:407:in
`join'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:395:in
`loop'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:395:in
`join'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:29:in
`run'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/bin/unicorn_rails:210
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/bin/unicorn_rails:19:in
`load'
/data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/bin/unicorn_rails:19

I don't get any other exceptions.  I've found other people have this
"IOError: closed stream" error coupled with "Errno::EMFILE" but we
have plenty of file-descriptors left.

Here's an lsof dump of the worker:

COMMAND   PID    USER   FD   TYPE             DEVICE        SIZE    NODE NAME
ruby    15209 zendesk  cwd    DIR              252,0        4096
2940970 /data/zendesk/releases/alpha
ruby    15209 zendesk  rtd    DIR                8,1        4096       2 /
ruby    15209 zendesk  txt    REG                8,1     2831149
246513 /usr/local/rvm/rubies/ree-1.8.7-2010.02/bin/ruby
ruby    15209 zendesk  mem    REG                8,1       10192
146145 /usr/lib/gconv/ISO8859-1.so
ruby    15209 zendesk  mem    REG                8,1       22848
263831 /lib/libnss_dns-2.9.so
ruby    15209 zendesk  mem    REG                8,1        9760
260915 /lib/libnss_mdns4_minimal.so.2
ruby    15209 zendesk  mem    REG                8,1       43472
263939 /lib/libnss_nis-2.9.so
ruby    15209 zendesk  mem    REG                8,1       35632
263929 /lib/libnss_compat-2.9.so
ruby    15209 zendesk  mem    REG                8,1       51616
263934 /lib/libnss_files-2.9.so
ruby    15209 zendesk  mem    REG              252,0       35772
3377231 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/em-http-request-0.2.7/lib/em_buffer.so
ruby    15209 zendesk  mem    REG              252,0       45360
3377237 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/em-http-request-0.2.7/lib/http11_client.so
ruby    15209 zendesk  mem    REG              252,0     1175256
3377180 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/eventmachine-0.12.10/lib/rubyeventmachine.so
ruby    15209 zendesk  mem    REG                8,1       35184
155430 /usr/lib/sasl2/libntlm.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1       18672
155418 /usr/lib/sasl2/libanonymous.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1       18672
155421 /usr/lib/sasl2/libcrammd5.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1       18640
155433 /usr/lib/sasl2/libplain.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1       18640
155427 /usr/lib/sasl2/liblogin.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1     1297736
140958 /usr/lib/libdb-4.6.so
ruby    15209 zendesk  mem    REG                8,1       22616
155410 /usr/lib/sasl2/libsasldb.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1       52000
155424 /usr/lib/sasl2/libdigestmd5.so.2.0.22
ruby    15209 zendesk  mem    REG                8,1       88824
263949 /lib/libresolv-2.9.so
ruby    15209 zendesk  mem    REG                8,1      105760
139781 /usr/lib/libsasl2.so.2.0.22
ruby    15209 zendesk  mem    REG              252,0     4110920
3384825 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/memcached-0.20.1/lib/rlibmemcached.so
ruby    15209 zendesk  mem    REG                8,1       97168
263938 /lib/libnsl-2.9.so
ruby    15209 zendesk  mem    REG                8,1     2086600
145819 /usr/lib/libmysqlclient.so.15.0.0
ruby    15209 zendesk  mem    REG              252,0      169011
3385005 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/mysql-2.8.1/lib/mysql_api.so
ruby    15209 zendesk  mem    REG              252,0      150950
3401474 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/yajl-ruby-0.7.6/lib/yajl_ext.so
ruby    15209 zendesk  mem    REG              252,0       29311
3367472 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/SystemTimer-1.2/lib/system_timer_native.so
ruby    15209 zendesk  mem    REG                8,1       13144
260768 /lib/libgpg-error.so.0.3.0
ruby    15209 zendesk  mem    REG                8,1      422664
260765 /lib/libgcrypt.so.11.4.4
ruby    15209 zendesk  mem    REG                8,1     1425368
141140 /usr/lib/libxml2.so.2.6.32
ruby    15209 zendesk  mem    REG                8,1      233152
439848 /usr/lib/libxslt.so.1.1.24
ruby    15209 zendesk  mem    REG                8,1       80792
439847 /usr/lib/libexslt.so.0.8.13
ruby    15209 zendesk  mem    REG              252,0      417564
3385563 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/nokogiri-1.4.1/lib/nokogiri/nokogiri.so
ruby    15209 zendesk  mem    REG                8,1       14220
252540 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/digest/md5.so
ruby    15209 zendesk  mem    REG                8,1       42715
252537 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/racc/cparse.so
ruby    15209 zendesk  mem    REG              252,0      121676
3465790 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/json-1.4.6/ext/json/ext/json/ext/generator.so
ruby    15209 zendesk  mem    REG              252,0       52561
3465795 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/json-1.4.6/ext/json/ext/json/ext/parser.so
ruby    15209 zendesk  mem    REG              252,0       20227
3401531 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/fast_xs-0.7.3/lib/fast_xs.so
ruby    15209 zendesk  mem    REG                8,1       46499
255504 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/site_ruby/1.8/x86_64-linux/iconv.so
ruby    15209 zendesk  mem    REG                8,1      139431
247137 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/bigdecimal.so
ruby    15209 zendesk  mem    REG                8,1      433200
247152 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/nkf.so
ruby    15209 zendesk  mem    REG              252,0      106488
3531787 /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn_http.so
ruby    15209 zendesk  mem    REG                8,1       53645
247158 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/strscan.so
ruby    15209 zendesk  mem    REG                8,1      128208
247138 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/socket.so
ruby    15209 zendesk  mem    REG                8,1       12144
247141 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/fcntl.so
ruby    15209 zendesk  mem    REG                8,1      329504
260852 /lib/libssl.so.0.9.8
ruby    15209 zendesk  mem    REG                8,1     1002954
247156 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/openssl.so
ruby    15209 zendesk  mem    REG                8,1      110815
247146 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/zlib.so
ruby    15209 zendesk  mem    REG                8,1       35684
247140 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/digest.so
ruby    15209 zendesk  mem    REG                8,1       96768
260746 /lib/libz.so.1.2.3.3
ruby    15209 zendesk  mem    REG                8,1     1594824
260668 /lib/libcrypto.so.0.9.8
ruby    15209 zendesk  mem    REG                8,1       15204
252539 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/digest/sha1.so
ruby    15209 zendesk  mem    REG                8,1       12723
247159 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/rational.so
ruby    15209 zendesk  mem    REG                8,1      351320
247142 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/syck.so
ruby    15209 zendesk  mem    REG                8,1       57484
247154 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/stringio.so
ruby    15209 zendesk  mem    REG                8,1       29818
247150 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/etc.so
ruby    15209 zendesk  mem    REG                8,1       51878
247139 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/ruby/1.8/x86_64-linux/thread.so
ruby    15209 zendesk  mem    REG                8,1      130151
263933 /lib/libpthread-2.9.so
ruby    15209 zendesk  mem    REG                8,1       96560
260672 /lib/libgcc_s.so.1
ruby    15209 zendesk  mem    REG                8,1     1023448
139822 /usr/lib/libstdc++.so.6.0.10
ruby    15209 zendesk  mem    REG                8,1     1502512
263944 /lib/libc-2.9.so
ruby    15209 zendesk  mem    REG                8,1      542928
263942 /lib/libm-2.9.so
ruby    15209 zendesk  mem    REG                8,1       39112
263932 /lib/libcrypt-2.9.so
ruby    15209 zendesk  mem    REG                8,1       14608
263945 /lib/libdl-2.9.so
ruby    15209 zendesk  mem    REG                8,1       31656
263930 /lib/librt-2.9.so
ruby    15209 zendesk  mem    REG                8,1      728968
246427 /usr/local/rvm/rubies/ree-1.8.7-2010.02/lib/libtcmalloc_minimal.so.0.0.0
ruby    15209 zendesk  mem    REG                8,1      135680
263936 /lib/ld-2.9.so
ruby    15209 zendesk  mem    REG                8,1       26048
139155 /usr/lib/gconv/gconv-modules.cache
ruby    15209 zendesk    0r   CHR                1,3
3125 /dev/null
ruby    15209 zendesk    1w   REG              252,0           0
3015360 /data/zendesk/log/unicorn.stdout.log
ruby    15209 zendesk    2w   REG              252,0       24585
3015203 /data/zendesk/log/unicorn.stderr.log-20110110 (deleted)
ruby    15209 zendesk    3w   REG              252,0 61617590073
3015097 /data/zendesk/log/unicorn.log-20110110 (deleted)
ruby    15209 zendesk    5u  unix 0xffff88043bdf9080
49439 /var/run/zendesk/unicorn.sock
ruby    15209 zendesk    6w  FIFO                0,6             5651378 pipe
ruby    15209 zendesk    7w   REG              252,0           0
3015676 /data/zendesk/log/newrelic_agent.log
ruby    15209 zendesk    8w   REG              252,0    68330048
3015673 /data/zendesk/log/production.log
ruby    15209 zendesk    9w   REG              252,0           0
3015677 /data/zendesk/log/resque.log
ruby    15209 zendesk   10u  IPv4           41707462
TCP 192.168.100.148:49364->192.168.100.183:mysql (ESTABLISHED)
ruby    15209 zendesk   11w   REG              252,0           0
3015672 /data/zendesk/log/mail_processing_production.log
ruby    15209 zendesk   12u  IPv4              55517
TCP 192.168.100.148:60198->192.168.100.182:mysql (ESTABLISHED)
ruby    15209 zendesk   13u  IPv4              55521
TCP 192.168.100.148:42351->192.168.100.150:11211 (ESTABLISHED)
ruby    15209 zendesk   14u  IPv4              55523
TCP 192.168.100.148:52409->192.168.100.148:11211 (ESTABLISHED)
ruby    15209 zendesk   15u  IPv4              56105
TCP 192.168.100.148:58562->192.168.100.140:11211 (ESTABLISHED)
ruby    15209 zendesk   16u  IPv4              56108
TCP 192.168.100.148:50500->192.168.100.144:11211 (ESTABLISHED)
ruby    15209 zendesk   17u  IPv4              56110
TCP 192.168.100.148:46723->192.168.100.142:11211 (ESTABLISHED)
ruby    15209 zendesk   18u  IPv4              56137
TCP 192.168.100.148:49568->192.168.100.146:11211 (ESTABLISHED)
ruby    15209 zendesk   19u  IPv4            5845535
TCP 192.168.100.148:43836->192.168.100.183:mysql (ESTABLISHED)
ruby    15209 zendesk   20u  IPv4              63603
TCP 192.168.100.148:41990->192.168.100.149:6379 (ESTABLISHED)
ruby    15209 zendesk   21u   REG                8,1           0
90084 /tmp/0.961990208371916 (deleted)
ruby    15209 zendesk   22u  IPv4           10795792
TCP 192.168.100.148:34587->192.168.100.183:mysql (ESTABLISHED)
ruby    15209 zendesk   23u  IPv4           33013892
TCP 192.168.100.148:54382->192.168.100.183:mysql (ESTABLISHED)


I'm tempted to exit the worker when I get this exception N number of
times as a workaround; it tends to only happen when I'm sleeping =)

From normalperson at yhbt.net  Tue Mar 22 17:39:05 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Tue, 22 Mar 2011 14:39:05 -0700
Subject: unicorn 1.1.x never-ending listen loop IOError exceptions
In-Reply-To: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>
References: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>
Message-ID: <20110322213905.GA23207@dcvr.yhbt.net>

David Nghiem <david at zendesk.com> wrote:
> We run unicorn 1.1.5 behind nginx under Ruby Enterprise Edition
> 2011.02.  We're using UNIX sockets instead of TCP sockets.
> 
> Once in a while, one worker will throw exceptions in a loop, taking up
> lots of CPU and log space.  Killing the worker removes the problem.
> Other workers are not affected.
> 
> Here's the exception:
> 
> E, [2011-01-10T10:13:13.518795 #15209] ERROR -- : Unhandled listen
> loop exception #<IOError: closed stream>.
> E, [2011-01-10T10:13:13.518881 #15209] ERROR -- :
> /data/zendesk/releases/alpha/vendor/bundle/ruby/1.8/gems/unicorn-1.1.5/lib/unicorn.rb:732:in
> `select'
> 
> I don't get any other exceptions.  I've found other people have this
> "IOError: closed stream" error coupled with "Errno::EMFILE" but we
> have plenty of file-descriptors left.

No Errno::EBADF errors at all?

> Here's an lsof dump of the worker:
> 
> COMMAND   PID    USER   FD   TYPE             DEVICE        SIZE    NODE NAME

<snip>  It looks like log reopening failed, below:

> ruby    15209 zendesk    1w   REG              252,0           0
> 3015360 /data/zendesk/log/unicorn.stdout.log
> ruby    15209 zendesk    2w   REG              252,0       24585
> 3015203 /data/zendesk/log/unicorn.stderr.log-20110110 (deleted)
> ruby    15209 zendesk    3w   REG              252,0 61617590073
> 3015097 /data/zendesk/log/unicorn.log-20110110 (deleted)

Permissions issue during log rotation?  Are you doing "killall -USR1
unicorn" by any chance or sending a bunch of USR1 signals faster than
unicorn can process them?

Try just sending one USR1 to the master process and letting the master
signal the workers.

-- 
Eric Wong

From david at zendesk.com  Tue Mar 22 20:04:48 2011
From: david at zendesk.com (David Nghiem)
Date: Tue, 22 Mar 2011 17:04:48 -0700
Subject: unicorn 1.1.x never-ending listen loop IOError exceptions
In-Reply-To: <20110322213905.GA23207@dcvr.yhbt.net>
References: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>
	<20110322213905.GA23207@dcvr.yhbt.net>
Message-ID: <AANLkTim=xPJ6Mo9160FvCBzh+RwLnsiOkb51WAhJmmV0@mail.gmail.com>

On Tue, Mar 22, 2011 at 2:39 PM, Eric Wong <normalperson at yhbt.net> wrote:

>> I don't get any other exceptions. ?I've found other people have this
>> "IOError: closed stream" error coupled with "Errno::EMFILE" but we
>> have plenty of file-descriptors left.
>
> No Errno::EBADF errors at all?

I didn't check before or after the block of Errno::IOError errors,
there were enough IOError logs to fill up 40gb of disk.  I'll capture
the entire deleted log next time.


> <snip> ?It looks like log reopening failed, below:
>
>> ruby ? ?15209 zendesk ? ?1w ? REG ? ? ? ? ? ? ?252,0 ? ? ? ? ? 0
>> 3015360 /data/zendesk/log/unicorn.stdout.log
>> ruby ? ?15209 zendesk ? ?2w ? REG ? ? ? ? ? ? ?252,0 ? ? ? 24585
>> 3015203 /data/zendesk/log/unicorn.stderr.log-20110110 (deleted)
>> ruby ? ?15209 zendesk ? ?3w ? REG ? ? ? ? ? ? ?252,0 61617590073
>> 3015097 /data/zendesk/log/unicorn.log-20110110 (deleted)
>
> Permissions issue during log rotation? ?Are you doing "killall -USR1
> unicorn" by any chance or sending a bunch of USR1 signals faster than
> unicorn can process them?
>
> Try just sending one USR1 to the master process and letting the master
> signal the workers.

We rotate logs once a night using logrotate.  There's a post-rotate
hook that sends USR1 once to the master.   There are three log files
but logrotate is run with the sharedscripts option so only one hook
runs.  The log dirs and unicorn processes are run as the same non-root
user.

I'll run an lsof on the healthy workers next time to see if they have
the same deleted log file descriptors.

Are the IOError exceptions and failure to rotate logs related?  My
wild guess was this specific worker's unix socket got corrupted
somehow.

From normalperson at yhbt.net  Tue Mar 22 21:09:35 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Wed, 23 Mar 2011 01:09:35 +0000
Subject: unicorn 1.1.x never-ending listen loop IOError exceptions
In-Reply-To: <AANLkTim=xPJ6Mo9160FvCBzh+RwLnsiOkb51WAhJmmV0@mail.gmail.com>
References: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>
	<20110322213905.GA23207@dcvr.yhbt.net>
	<AANLkTim=xPJ6Mo9160FvCBzh+RwLnsiOkb51WAhJmmV0@mail.gmail.com>
Message-ID: <20110323010935.GA25934@dcvr.yhbt.net>

David Nghiem <david at zendesk.com> wrote:
> On Tue, Mar 22, 2011 at 2:39 PM, Eric Wong <normalperson at yhbt.net> wrote:
> > No Errno::EBADF errors at all?
> 
> I didn't check before or after the block of Errno::IOError errors,
> there were enough IOError logs to fill up 40gb of disk.  I'll capture
> the entire deleted log next time.

Yikes.  The first few lines of error messages should be enough.

> >> ruby ? ?15209 zendesk ? ?1w ? REG ? ? ? ? ? ? ?252,0 ? ? ? ? ? 0
> >> 3015360 /data/zendesk/log/unicorn.stdout.log
> >> ruby ? ?15209 zendesk ? ?2w ? REG ? ? ? ? ? ? ?252,0 ? ? ? 24585
> >> 3015203 /data/zendesk/log/unicorn.stderr.log-20110110 (deleted)
> >> ruby ? ?15209 zendesk ? ?3w ? REG ? ? ? ? ? ? ?252,0 61617590073
> >> 3015097 /data/zendesk/log/unicorn.log-20110110 (deleted)
> >
> > Permissions issue during log rotation? ?Are you doing "killall -USR1
> > unicorn" by any chance or sending a bunch of USR1 signals faster than
> > unicorn can process them?
> >
> > Try just sending one USR1 to the master process and letting the master
> > signal the workers.
> 
> Are the IOError exceptions and failure to rotate logs related?  My
> wild guess was this specific worker's unix socket got corrupted
> somehow.

There's a good possibility.  The USR1 signal closes a pipe we normally
select() on.  The closed pipe triggers a wakeup from select(), which
triggers the log reopening (and reopening of the pipe).

Do you see:  "worker=$NUM done reopening logs" messages in your
logs?

Another explanation is somehow you have a library that does something
along the lines of:

 ObjectSpace.each_object(IO) { |io| io.close unless io.closed? }

-- 
Eric Wong

From list at mrtech.ru  Wed Mar 23 14:12:13 2011
From: list at mrtech.ru (Troex Nevelin)
Date: Wed, 23 Mar 2011 21:12:13 +0300
Subject: [ANN] raindrops updates (mainly for Linux users)
In-Reply-To: <20110318180902.GA6076@dcvr.yhbt.net>
References: <20110318180902.GA6076@dcvr.yhbt.net>
Message-ID: <4D8A37FD.6010807@mrtech.ru>

On 03/18/2011 09:09 PM, Eric Wong wrote:
> I figured I should cross post here.  raindrops (listen queue stats
> package for Unicorn/Rainbows!) has gotten some largish updates over the
> past few weeks:
>
> raindrops 0.5.0 release announcement:
>    http://mid.gmane.org/20110317033547.GA30653 at dcvr.yhbt.net
>
> demo site changes (will probably be in 0.6.0 soon):
>    http://mid.gmane.org/20110318094637.GA3526 at dcvr.yhbt.net
>
> Feedback would be greatly appreciated (either here or
> raindrops at librelist.com).  There are no backwards-incompatible changes
> for existing users as far as I'm aware of.

This update is really great! Now we can see 'Mean' users awaiting for 
their requests to be served - this information is very useful, because 
we now can calculate avarage wait time for each request.

For example we know that an average request time is 0.15 sec, than if 
our mean queue is 32 requests, each request will be served in 
(32+1)*0.15=4.95 seconds. This information is very useful and even 
NewRelics cannot provide it.

I'm going to build graphs for this data. You can see live stats for 700k 
request per day project running on 2x Xeon E5410 2.33 Ghz (4 cores each):

http://kinokopilka.tv:9292/queued/%2Fvar%2Fwww%2Fkk%2Fcurrent%2Ftmp%2Fsockets%2Funicorn.sock.html

From list at mrtech.ru  Wed Mar 23 14:25:58 2011
From: list at mrtech.ru (Troex Nevelin)
Date: Wed, 23 Mar 2011 21:25:58 +0300
Subject: Memcache reset on rails 2.3 with preload
In-Reply-To: <4D794A34.4090900@gmail.com>
References: <4D790788.8010305@mrtech.ru> <4D794A34.4090900@gmail.com>
Message-ID: <4D8A3B36.4000600@mrtech.ru>

On 03/11/2011 01:01 AM, Lawrence Pit wrote:
> I use:
>
> if Rails.cache.is_a?(ActiveSupport::Cache::MemCacheStore)
> Rails.cache.instance_variable_get(:@data).reset
> end

Thanks a lot, that is exactly what I needed

From list at mrtech.ru  Wed Mar 23 15:27:56 2011
From: list at mrtech.ru (Troex Nevelin)
Date: Wed, 23 Mar 2011 22:27:56 +0300
Subject: [ANN] raindrops updates (mainly for Linux users)
In-Reply-To: <4D8A37FD.6010807@mrtech.ru>
References: <20110318180902.GA6076@dcvr.yhbt.net> <4D8A37FD.6010807@mrtech.ru>
Message-ID: <4D8A49BC.1080903@mrtech.ru>

On 03/23/2011 09:12 PM, Troex Nevelin wrote:
> For example we know that an average response time is 0.15 sec, than if 
> our mean queue is 32 requests, each request will be served in 
> (32+1)*0.15=4.95 seconds.

Probably it should be:
((mean_queue + 1) * avg_response_time) / workers 

so in my example it would be:
((32 + 1) * 0.15) / 8 =~ 0.62 sec

From normalperson at yhbt.net  Wed Mar 23 18:26:29 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Wed, 23 Mar 2011 15:26:29 -0700
Subject: [ANN] raindrops updates (mainly for Linux users)
In-Reply-To: <4D8A37FD.6010807@mrtech.ru>
References: <20110318180902.GA6076@dcvr.yhbt.net> <4D8A37FD.6010807@mrtech.ru>
Message-ID: <20110323222629.GB26826@dcvr.yhbt.net>

Troex Nevelin <list at mrtech.ru> wrote:
> This update is really great!

Awesome, your feedback is much appreciated.   I'll forward a link to
your posts to the raindrops list.

-- 
Eric Wong

From normalperson at yhbt.net  Tue Mar 29 12:24:44 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Tue, 29 Mar 2011 09:24:44 -0700
Subject: unicorn 1.1.x never-ending listen loop IOError exceptions
In-Reply-To: <AANLkTim3xZMKxcmQzPjcB7vrJKiHoc2f8WoU8RatrDum@mail.gmail.com>
References: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>
	<20110322213905.GA23207@dcvr.yhbt.net>
	<AANLkTim=xPJ6Mo9160FvCBzh+RwLnsiOkb51WAhJmmV0@mail.gmail.com>
	<20110323010935.GA25934@dcvr.yhbt.net>
	<AANLkTim3xZMKxcmQzPjcB7vrJKiHoc2f8WoU8RatrDum@mail.gmail.com>
Message-ID: <20110329162444.GA12293@dcvr.yhbt.net>

David Nghiem <david at zendesk.com> wrote:
> I'm wondering if the problem is three logs for unicorn (unicorn.log, and the
> stdout, stderr logs) are rotated by logrotate, but the first log triggers
> the USR1 and in the middle of rotating the remaining logs, the unicorn log
> reopening gets confused.
> 
> I'll make it so the USR1 signal is sent after all three logs have been
> rotated.   Let this run for a couple months and report back.

That could be it.   logrotate (without the scary/expensive
"copytruncate" option) works very quickly for individual files.  Do you
have the "sharedscripts" option in logrotate enabled?  You should enable
"sharedscripts"

I should post a logrotate example config so other users don't run into
the same problem...

> Thanks, Eric, for your help.  I've been running unicorn since the pre-1.0
> days and it's my favorite ruby software.

Good to know!

-- 
Eric Wong

From normalperson at yhbt.net  Tue Mar 29 12:54:58 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Tue, 29 Mar 2011 09:54:58 -0700
Subject: [PATCH] add examples/logrotate.conf
In-Reply-To: <20110329162444.GA12293@dcvr.yhbt.net>
References: <AANLkTin18jkoaMkj-kZf-Seqw8hz8aEQ0bZqL_Jf79xo@mail.gmail.com>
	<20110322213905.GA23207@dcvr.yhbt.net>
	<AANLkTim=xPJ6Mo9160FvCBzh+RwLnsiOkb51WAhJmmV0@mail.gmail.com>
	<20110323010935.GA25934@dcvr.yhbt.net>
	<AANLkTim3xZMKxcmQzPjcB7vrJKiHoc2f8WoU8RatrDum@mail.gmail.com>
	<20110329162444.GA12293@dcvr.yhbt.net>
Message-ID: <20110329165458.GA14389@dcvr.yhbt.net>

Eric Wong <normalperson at yhbt.net> wrote:
> That could be it.   logrotate (without the scary/expensive
> "copytruncate" option) works very quickly for individual files.  Do you
> have the "sharedscripts" option in logrotate enabled?  You should enable
> "sharedscripts"

I actually went with lastaction in my example (I keep forgetting it over
the years and going to postrotate + sharedscripts for some reason..)

> I should post a logrotate example config so other users don't run into
> the same problem...

http://unicorn.bogomips.org/examples/logrotate.conf and pushed
this out to unicorn.git

>From c1ebb313735a280582d87c1ba44619aa47e00b06 Mon Sep 17 00:00:00 2001
From: Eric Wong <normalperson at yhbt.net>
Date: Tue, 29 Mar 2011 09:47:26 -0700
Subject: [PATCH] add examples/logrotate.conf

logrotate is the de facto tool for logrotation, so an
example config for highlighting important parts are in order.

Since our USR1 signal handling is part of the crusade against
the slow and lossy "copytruncate" option, be sure to
emphasize that :)
---
 examples/logrotate.conf |   29 +++++++++++++++++++++++++++++
 1 files changed, 29 insertions(+), 0 deletions(-)
 create mode 100644 examples/logrotate.conf

diff --git a/examples/logrotate.conf b/examples/logrotate.conf
new file mode 100644
index 0000000..03fefc6
--- /dev/null
+++ b/examples/logrotate.conf
@@ -0,0 +1,29 @@
+# example logrotate config file, I usually keep this in
+# /etc/logrotate.d/unicorn_app on my Debian systems
+#
+# See the logrotate(8) manpage for more information:
+#    http://linux.die.net/man/8/logrotate
+
+# Modify the following glob to match the logfiles your app writes to:
+/var/log/unicorn_app/*.log {
+	# this first block is mostly just personal preference, though
+	# I wish logrotate offered an "hourly" option...
+	daily
+	missingok
+	rotate 180
+	compress # must use with delaycompress below
+	dateext
+
+	# this is important if using "compress" since we need to call
+	# the "lastaction" script below before compressing:
+	delaycompress
+
+	# note the lack of the evil "copytruncate" option in this
+	# config.  Unicorn supports the USR1 signal and we send it
+	# as our "lastaction" action:
+	lastaction
+		# assuming your pid file is in /var/run/unicorn_app/pid
+		pid=/var/run/unicorn_app/pid
+		test -s $pid && kill -USR1 "$(cat $pid)"
+	endscript
+}
-- 
Eric Wong

From list at mrtech.ru  Wed Mar 30 07:48:21 2011
From: list at mrtech.ru (Troex Nevelin)
Date: Wed, 30 Mar 2011 15:48:21 +0400
Subject: TAG for process list view and start-up-scripts
Message-ID: <4D931885.9070605@mrtech.ru>

1. Thin has 'tag' option in config, it is shown process list.
This feature was very useful for me, because on servers I
run 2 and more unicorn instances it would be nice to identify
them fast with one look.


2. It would be nice to have example sh startup scripts come with
unicorn. I know they can be very specific but anyway it will help
people start using unicorn.

From normalperson at yhbt.net  Wed Mar 30 13:52:39 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Wed, 30 Mar 2011 10:52:39 -0700
Subject: TAG for process list view and start-up-scripts
In-Reply-To: <4D931885.9070605@mrtech.ru>
References: <4D931885.9070605@mrtech.ru>
Message-ID: <20110330175239.GA9877@dcvr.yhbt.net>

Troex Nevelin <list at mrtech.ru> wrote:
> 1. Thin has 'tag' option in config, it is shown process list.
> This feature was very useful for me, because on servers I
> run 2 and more unicorn instances it would be nice to identify
> them fast with one look.

I won't be supporting something like this because everybody has
a different idea of what it should be and it bloats terminals.
A similar idea that was rejected in this thread:
  http://mid.gmane.org/20100702205004.GI4248 at jeremyevans.local

In a related thread, Jeremy's monkey patch (which should work for all
future/current versions):
  http://mid.gmane.org/AANLkTimQoEH7QBPn9ppMP-W_kWPi5UyhKwm4jMmSvHlu at mail.gmail.com

I only do the "master"/"worker" distinction because it's what
nginx users are used to seeing and it can usually fit in argv[0]
without needing more space (moving + clobbering environ)[1]

> 2. It would be nice to have example sh startup scripts come with
> unicorn. I know they can be very specific but anyway it will help
> people start using unicorn.

Something like examples/init.sh in our source tree?
All of the examples/ directories is uploaded to
http://unicorn.bogomips.org/examples/ too

[1] /proc/$PID/environ under Linux is a sometimes useful debugging tool
    but broken under Ruby 1.9.2, I'm working on fixing this for 1.9.3
-- 
Eric Wong

From jheiss at aput.net  Wed Mar 30 14:19:48 2011
From: jheiss at aput.net (Jason Heiss)
Date: Wed, 30 Mar 2011 11:19:48 -0700
Subject: Does unicorn support mount the webapp with a prefix like Thin
	does?
In-Reply-To: <AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>
References: <AANLkTinDS86owzdni8bgxN4RE=B=ocXAsqzcdH1u93U2@mail.gmail.com>
	<AANLkTikG120HJOyFGWE3NPWLAPf81-_D6=e=Ai8=PATS@mail.gmail.com>
Message-ID: <90F62515-83A6-4D2F-ABD8-7B60476B9BF6@aput.net>

On Mar 21, 2011, at 12:37 AM, Guang-Nan Cheng wrote:

> For example, I have a webapp that handles /welcome.
> 
> Does unicorn support to mount the entire webapp at /prefix so request
> to /prefix/welcome are handled?

If it is a Rails app you can put the following in your unicorn config file:

ENV['RAILS_RELATIVE_URL_ROOT'] = '/prefix'


From mkocher at pivotallabs.com  Thu Mar 31 13:45:32 2011
From: mkocher at pivotallabs.com (Matthew Kocher)
Date: Thu, 31 Mar 2011 10:45:32 -0700
Subject: Reopening logs - why no relative paths?
Message-ID: <AANLkTimM0OV2=kRKjpB0Bj1AFMTpiwMj5H223-45fb6P@mail.gmail.com>

We just spend a couple hours trying to figure out why sending a USR1
signal wasn't causing unicorn to reopen logs.  Looking at the source
reveals that only files opened with absolute paths are eligible for
rotation, the others are silently skipped.

We use the fairly standard practice of always logging to
log/unicorn.error.log and creating symlinks to where we'd like the
logs to live on individual hosts.  We'll change our configs for the
time being, but is there a reason to not reopen the relative paths?

Thanks,
Matthew

From normalperson at yhbt.net  Thu Mar 31 18:02:48 2011
From: normalperson at yhbt.net (Eric Wong)
Date: Thu, 31 Mar 2011 15:02:48 -0700
Subject: Reopening logs - why no relative paths?
In-Reply-To: <AANLkTimM0OV2=kRKjpB0Bj1AFMTpiwMj5H223-45fb6P@mail.gmail.com>
References: <AANLkTimM0OV2=kRKjpB0Bj1AFMTpiwMj5H223-45fb6P@mail.gmail.com>
Message-ID: <20110331220248.GA1826@dcvr.yhbt.net>

Matthew Kocher <mkocher at pivotallabs.com> wrote:
> We just spend a couple hours trying to figure out why sending a USR1
> signal wasn't causing unicorn to reopen logs.  Looking at the source
> reveals that only files opened with absolute paths are eligible for
> rotation, the others are silently skipped.
> 
> We use the fairly standard practice of always logging to
> log/unicorn.error.log and creating symlinks to where we'd like the
> logs to live on individual hosts.  We'll change our configs for the
> time being, but is there a reason to not reopen the relative paths?

It's ambiguous due to directory changes[1].  It may be easier now that
we have the "working_directory" directive now, but configuration changes
also change (or unset) the "working_directory" which adds to the
confusion.

We could try allowing it and see what happens, I suppose...


[1] - in my experience with writing daemons, directory changes make
      everything much harder all around.
-- 
Eric Wong