Re: "EINVAL" Crashing with Instant Rails and PostgreSQL on Windows

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re: "EINVAL" Crashing with Instant Rails and PostgreSQL on Windows

Jeff Casimir
Zed,

I've found a little more info about this "EINVAL" crashes I'm having.  
 From scgi.log:

[ERR][508] C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:259:in `close'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:205:in `join'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:205:in `collect_thread'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:181:in `listen'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:179:in `initialize'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:179:in `new'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:179:in `listen'
C:/ruby/irails/ruby/lib/ruby/site_ruby/1.8/scgi.rb:412:in `run'
C:/ruby/irails/ruby/bin/scgi_service:61

I have that repeated many times.  The consequence is that the server
spawns new ruby processes until it chokes.

Any thoughts?  Can I give you more information of any kind?

-Jeff

Zed A. Shaw wrote:

>Hey Jeff,
>
>That actually looks like the socket being used has become invalid.  The
>man page on my Linux box says:
>
>EINVAL fd is attached to an object which is unsuitable for
>writing;  or the  file  was  opened  with  the  O_DIRECT flag, and
>either the address specified in buf, the value specified in count,  or
>the current file offset is not suitably aligned.
>
>So that means your connection is getting closed or aborted in the
>middle of a write operation.  Not sure if this helps though on Win32.
>
>Zed
>
>On Tue, 13 Dec 2005 09:58:20 -0500
>Jeff Casimir <[hidden email]> wrote:
>
>  
>
>>Hey all,
>>
>>I was getting some similar crashes when using webbrick with high (a
>>few request every second or so) load so I moved up to Apache.
>>Specifically, I am using the "InstantRails" package with Apache/SCGI
>>and PostgreSQL instead of mySQL.  However the same problem has sprung
>>up again and causes the server to spawn hundreds of separate ruby.exe
>>processes.  Is the below stack trace meaningful to anyone?  I don't
>>_think_ this is a problem with my app, it looks like a crash in the
>>postgres module. Thoughts?
>>
>>Thanks,
>>Jeff
>>
>>    
>>
>_______________________________________________
>Rails mailing list
>[hidden email]
>http://lists.rubyonrails.org/mailman/listinfo/rails
>  
>

_______________________________________________
Rails mailing list
[hidden email]
http://lists.rubyonrails.org/mailman/listinfo/rails