librelist archives

« back to archive

[ERROR] - Bad file descriptor?

[ERROR] - Bad file descriptor?

From:
Matt Towers
Date:
2011-03-10 @ 16:29
Any ideas what might be causing this error to get logged when a SIGTERM or
SIGINT is sent to m2?

[ERROR] (src/task/net.c:236: errno: Bad file descriptor) Failed to accept 
after doing a poll on the socket.

✈ Matt


Re: [mongrel2] [ERROR] - Bad file descriptor?

From:
Zed A. Shaw
Date:
2011-03-10 @ 19:02
On Thu, Mar 10, 2011 at 08:29:46AM -0800, Matt Towers wrote:
> Any ideas what might be causing this error to get logged when a SIGTERM 
or SIGINT is sent to m2?
> 
> [ERROR] (src/task/net.c:236: errno: Bad file descriptor) Failed to 
accept after doing a poll on the socket.

Looks like the signal hit right when the accept on the socket happened.
It's a rare thing but shouldn't cause any problems.  
Did the server crash after that?

-- 
Zed A. Shaw
http://zedshaw.com/

Re: [mongrel2] [ERROR] - Bad file descriptor?

From:
Matt Towers
Date:
2011-03-10 @ 22:17
It didn't crash but it continues to dump the following to the error.log 
file over the course of about 10 seconds.  In this case, there are no 
handlers registered in config.sqlite. 

[INFO] (src/mongrel2.c:83) SHUTDOWN REQUESTED: GRACEFUL (SIGINT again to EXIT NOW)
[ERROR] (src/task/net.c:236: errno: Bad file descriptor) Failed to accept 
after doing a poll on the socket.
[INFO] (src/mongrel2.c:312) Shutdown requested, goodbye.
[INFO] (src/mongrel2.c:252) Waiting for connections to die: 1
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:256) Waiting for connections to die: 2
[INFO] (src/mongrel2.c:261) Removing pid file /run/mongrel2.pid

✈ Matt



On Mar 10, 2011, at 11:02 , Zed A. Shaw wrote:

> Looks like the signal hit right when the accept on the socket happened.
> It's a rare thing but shouldn't cause any problems.  
> Did the server crash after that?