ringleader hanging

[danrabinowitz]

Ringleader's been hanging on me frequently for the past few weeks.

When I try to stop an app, or when stops after idle_timeout, it freezes at this line of code: https://github.com/aniero/ringleader/blob/master/lib/ringleader/process.rb#L195

CTRL-c doesn't cancel. I have to kill -9 ringleader.

It behaved fine before (roughly) this commit:
https://github.com/aniero/ringleader/commit/27639b4e9972ceba7a7a8d148eed45eb1c3a9043

Anyone else having this issue?

[MikeSilvis]

i noticed ringleader hanging and even with:

kill_with: INT

It still behaves oddly forcing me to kill -9 my apps individually to get them to start back up in ringleader.

[zerowidth]

This has been discussed in the (closed, un-reopenable) PR #22 . The latest alpha release (1.1.5.pre0) locks sys-proctable to 1.9.1, which should be ok.

cc @tapajos

[zerowidth]

oh also cc @wengzilla

[tapajos]

@aniero 1.1.5.pre0 is working well here.

[danrabinowitz]

Here too.

[tapajos]

@aniero,

Using this new version I'm getting this error very offen:

ERROR 10:56:01.045796 | Ringleader::App crashed!
Errno::EMFILE: Too many open files - pipe
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-io-0.12.1/lib/celluloid/io/reactor.rb:16:in `new'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-io-0.12.1/lib/celluloid/io/reactor.rb:16:in `initialize'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-io-0.12.1/lib/celluloid/io/mailbox.rb:10:in `new'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-io-0.12.1/lib/celluloid/io/mailbox.rb:10:in `initialize'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid.rb:171:in `new'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid.rb:171:in `block in mailbox_class'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid.rb:204:in `call'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid.rb:204:in `mailbox_factory'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid.rb:225:in `actor_options'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid.rb:100:in `new'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/ringleader-1.1.5.pre0/lib/ringleader/app.rb:96:in `proxy_to_app'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid/calls.rb:53:in `public_send'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid/calls.rb:53:in `dispatch'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid/actor.rb:327:in `block in handle_message'
/Users/tapajos/.rvm/gems/ruby-1.9.3-p392/gems/celluloid-0.12.4/lib/celluloid/tasks/task_fiber.rb:24:in `block in initialize'

[zerowidth]

@tapajos Looks like you're not on the latest. Celluloid 0.13.0 is the latest dependency, and I also changed where the socket proxying happens so only one IO reactor per app is used, instead of one reactor per connection. This should drastically cut down on the number of file handles ringleader uses.

[tapajos]

@aniero,

0.13.0 is not allowed. check: https://github.com/aniero/ringleader/blob/master/ringleader.gemspec#L19

[zerowidth]

I tell you what really helps: pushing code to github after writing it. Try now...

[tapajos]

@aniero ha ha ha

[tapajos]

I will test it now!!!

[tapajos]

@aniero I generated a new version locally and seems it is ok now. Can you release a new beta?

[zerowidth]

I'll release a new beta once I test celluloid 0.13.1!

[zerowidth]

ringleader 1.1.5.pre1 is released.

The hanging bug should be fixed now, as is an error we found in celluloid re: unhandled and un-rescuable IO errors with prematurely-closed sockets.

There may still be an issue with reel, per celluloid/reel#33, but it's been running fine for me. Let me know if there are any problems, either with proxying or with the web interface.