Re: [dev] orphan wmiirc's, wmii's / ruby wmiirc

I started with the ruby wmiirc shipped in 3.9b1, but have been using
the github repo since my latest launch (with repo updates while
running).

So, I decided to experiment with some cleanup, killing the processes I
was pretty sure didn't belong anymore, and seeing the results.

1) kill 18600
* wmii
* my current wmii still running
* xmessage error popped up:
#<Errno::ECONNRESET: Connection reset by peer
Ensure that (1) the WMII_ADDRESS environment variable is set and that (2)
it correctly specifies the filesystem path of wmii's IXP socket file,
which is typically located at "/tmp/ns.$USER.:$DISPLAY/wmii".
>
/usr/lib64/ruby/gems/1.8/gems/rumai-3.2.0/lib/rumai/ixp/message.rb:38:in `read'
...

2) kill 17775
* ruby wmiirc
* my current wmii is NOT running
* saved by ~/.wmii/wmiirc
* xmessage error popped up:
#<SignalException: SIGTERM>
/usr/lib64/ruby/gems/1.8/gems/rumai-3.2.0/lib/rumai/ixp/message.rb:38:in `read'
...

3) kill 19328
* ruby wmiirc
* my current wmii still running
* xmessage error popped up -- similar to above
* In the this case, I clicked 'Ignore', so wmiirc wouldnt' be
relaunched; everything wmiirc related AOK now (i.e only 1).

4) kill 2447
* wmii
* nothing popped up
* session still working

5) kill 19866
* wmii
* nothing popped up
* session still working

At this point, I have 1 wmii and 1 wmiirc. Successful launches of
~/.wmii/wmiirc do relaunch (pid changes), and I still have only 1
instance overall.

I'll keep an eye out when I next log out of X. My ~/.xinitrc ends
with "until wmii; do :; done", so maybe the problem of non-dying
wmii's lies there.

Thanks for the points and info.
__armando

On Thu, Dec 10, 2009 at 11:15 PM, Suraj Kurapati <sunaku_AT_gmail.com> wrote:
> Hi,
>
> On Thu, Dec 10, 2009 at 12:31 PM, Armando Di Cianno
> <armando_AT_goodship.net> wrote:
>> I'm using the sunaku wmiirc setup, with some modifications, from
>> github.  I tend to restart/re-run wmiirc a lot, as I make
>> modifications. [...] I have multiple wmiirc's (ruby ones, in this case),
>> but only one seems germane to the current session.
>
> Perhaps your wmiirc fork is from before this commit?
> (at least, when those now-zombie processes were run)
>
> http://github.com/sunaku/wmiirc/commit/3d642a5b216fe246d5e9db9896d9a592fec2e22e
>
> I am unable to reproduce your issue locally.  The old wmiirc cleanly
> exits when it sees the handshake string on /event/ every time.  And
> the reload action is causing the wmiirc to exec() within the same
> process, with no additional zombies to be seen.
>
> Hope this helps.
>
>
Received on Fri Dec 11 2009 - 19:54:16 UTC