Backport #1484: Ruby 1.8.6_p368 and Ruby 1.8.7_p160 have threading regressions

• Status: Closed
• Assignee: -

[hansdegraaff (Hans de Graaff)]

=begin
Changes introduced in ruby 1.8.6_368 and ruby 1.8.7_p160 have regressions in the way they handle threading.

The following bug report for Capistrano details the regression: https://capistrano.lighthouseapp.com/projects/8716/tickets/79-capistrano-hangs-on-shell-command-for-many-computers-on-ruby-186-p368#ticket-79-22

Basically, when trying to deploy to more than one machine capistrano now hangs while eating up all CPU, instead of actually running the deployment tasks. Deploying to a single machine or deploying with e.g. ruby 1.8.7_p72 works fine.
=end

[rogerdpack (Roger Pack)]

=begin

Basically, when trying to deploy to more than one machine capistrano now hangs while eating up all CPU, instead of actually running the deployment tasks. Deploying to a single machine or deploying with e.g. ruby 1.8.7_p72 works fine.

Is this still the case?
=r
=end

[luislavena (Luis Lavena)]

=begin
On Wed, Jul 8, 2009 at 5:24 PM, Roger [e-mail] wrote:

Issue #1484 has been updated by Roger Pack.

Basically, when trying to deploy to more than one machine capistrano now hangs while eating up all CPU, instead of actually running the deployment tasks. Deploying to a single machine or deploying with e.g. ruby 1.8.7_p72 works fine.

Is this still the case?
=r

I see similar problems with all threading related tasks on Mongrel and
sqlite3-ruby threaded tests.

Could be a duplicate of the other bug report.

Luis Lavena
AREA 17

Perfection in design is achieved not when there is nothing more to add,
but rather when there is nothing more to take away.
Antoine de Saint-Exupéry

=end

[hansdegraaff (Hans de Graaff)]

=begin
http://redmine.ruby-lang.org/issues/show/1471 contains some discussion and patches that I thought were successfull, but it seems in my initial testing I was lucky or not careful enough. In any case I'm still seeing this problem with capistrano deployments when deploying to more than one machine at the time.
=end

[dazuma (Daniel Azuma)]

=begin
I just reported http://redmine.ruby-lang.org/issues/show/1993 against 1.8.7, which looks like it's the same issue. (Sorry, I should have looked through the bug lists of the other versions-- I had thought this was 1.8.7 specific.) That other report has some analysis of rb_thread_schedule() that I think explains why this is happening. See also http://www.daniel-azuma.com/blog/view/z2ysbx0e4c3it9/ruby_1_8_7_io_select_threading_bug
=end

[hongli (Hongli Lai)]

=begin
Hi guys, any updates on this ticket? Has it been fixed yet?
=end

[hansdegraaff (Hans de Graaff)]

=begin
As far as I know this has not been fixed yet, but net-ssh-2.0.14 contains a workaround that works with capistrano.
=end

[jeremyevans0 (Jeremy Evans)]

• Tracker changed from Bug to Backport
• Project changed from 2 to 5
• Description updated (diff)
• Status changed from Open to Closed
• ruby -v deleted (1.8.6_p368, ruby 1.8.7_p160)