Feature #4243: [PATCH] DRb should not forefully close connections on DRb#stop_service

• Status: Closed
• Assignee: seki (Masatoshi Seki)
• Target version: 2.0.0

[zimbatm (zimba tm)]

=begin
When DRb#stop_service is called, all connections get terminated by
calling Thread.kill on them. It means that all code that is invoked trough
DRb must be exception-safe. This poses the same kind of issues we have
with Kernel#timeout.

This patch changes the behavior or DRb to only kill the accepting
thread, so that no new connections are made. New queries are also
rejected. DRb#stop_service is synchronous and waits for all child-threads
to terminate.

Please note that this patch is quite old, so I may have forgotten some details
of it. It was posted on rubyforge at http://rubyforge.org/tracker/?func=detail&atid=1698&aid=14438&group_id=426.
The new version got imported into git and applies on top of ae3666b38fd723231c0bbc3c33f4603b6e0e0ef0 or svn r30363 (not sure about the patch format)

This patch can easily be back-ported to 1.8.x since drb.rb hasn't changed much.

Cheers,
zimbatm
=end

[usa (Usaku NAKAMURA)]

• Status changed from Open to Assigned
• Assignee set to seki (Masatoshi Seki)

=begin

=end

[nahi (Hiroshi Nakamura)]

• Target version changed from 1.9.2 to 1.9.3

[mame (Yusuke Endoh)]

• Tracker changed from Bug to Feature

[mame (Yusuke Endoh)]

This ticket is a feature, not a bug. Moving to Feature tracker.
And the feature proposed introduces incompatibility. We have to
be careful about the ticket.

--
Yusuke Endoh [e-mail]

[mame (Yusuke Endoh)]

• Target version changed from 1.9.3 to 2.0.0

[ko1 (Koichi Sasada)]

• Status changed from Assigned to Closed

I close this ticket. Seki-san said no such problem now.
Please re-open this ticket if the issue is not gone.