Closed GoogleCodeExporter closed 9 years ago
Which mode are you using pywebsocket in? standalone or with Apache.
Could you please also give us stack trace or error log around the segmentation
fault if possible? If you're using it with Apache, you can get more detailed
log from pywebsocket by changing LogLevel of Apache to debug.
Original comment by tyoshino@chromium.org
on 30 Jan 2012 at 8:20
We are using with Apache.
Original comment by sdid...@gmail.com
on 30 Jan 2012 at 2:58
Stack trace to follow.
Original comment by sdid...@gmail.com
on 30 Jan 2012 at 2:59
Stack information:
apache error log (debug level):
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake: Client's opening handshake resource:
'/wss/mustang/client'
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake: Client's opening handshake headers: {'Origin':
'https://bmw2', 'Upgrade': 'websocket', 'Sec-WebSocket-Version': '13', 'Host':
'bmw2', 'Sec-WebSocket-Key': 'fL8MtY99Eb0es6DBZoLhZg==', 'Connection':
'Upgrade'}
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake: Trying IETF HyBi latest protocol
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake.hybi.Handshaker: Sec-WebSocket-Key:
'fL8MtY99Eb0es6DBZoLhZg==' (7c bf 0c b5 8f 7d 11 bd 1e b3 a0 c1 66 82 e1 66)
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake.hybi.Handshaker: Sec-WebSocket-Accept:
'ZYQIvHVYArAA4ZV7pR6tvln1Om4=' (65 84 08 bc 75 58 02 b0 00 e1 95 7b a5 1e ad be
59 f5 3a 6e)
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake.hybi.Handshaker: IETF HyBi protocol
[Mon Jan 30 08:06:14 2012] [debug] _apachemodule.c(65):
mod_pywebsocket.handshake.hybi.Handshaker: Sent server's opening handshake:
'HTTP/1.1 101 Switching Protocols\\r\\nUpgrade: websocket\\r\\nConnection:
Upgrade\\r\\nSec-WebSocket-Accept: ZYQIvHVYArAA4ZV7pR6tvln1Om4=\\r\\n\\r\\n'
[Mon Jan 30 08:06:14 2012] [info] mod_pywebsocket.handshake: Established (IETF
HyBi latest protocol)
[Mon Jan 30 08:06:15 2012] [notice] child pid 14612 exit signal Segmentation
fault (11)
Our log has lines like:
######Made it to parse the packet
######Session ID is: C7D9EC1B-3D2C-416C-8472-09BE8BBCE6F2
######RU: Got the session id: 11
######on to make the thread
###1###ready to make thread
###starting a queue for :client_queue_11
######thread made, name is: client_queue_11
######thread name inserted into DB sql was: UPDATE `Session` SET
`WSQueueTableName` = 'client_queue_11' WHERE `User_ID` = '11';
###queue_out_to_client: User_ID is: 11 sending message:
######Got an empty line, this will cause a socket close
### !!!!!!!! Send out message from thread failed:
The first two lines of our log indicate that the socket was open and the client
sent his sessionID to the server with no problem.
Original comment by sdid...@gmail.com
on 30 Jan 2012 at 3:15
Firefox updated and now it also fails.
Original comment by sdid...@gmail.com
on 7 Feb 2012 at 10:33
Hmm, it works here with Apache. If possible, could you please tell me the
interaction with mod_pywebsocket which caused segfault on your environment?
Maybe, you receive a packet by receive_message() call and parse the returned
value into
C7D9EC1B-3D2C-416C-8472-09BE8BBCE6F2 and something. Then you called send_message() with some argument maybe containing 11.
That's all?
Original comment by tyoshino@chromium.org
on 8 Feb 2012 at 4:05
It appears that we may have a problem with our ssl certificate on the target
machine. With https we can just tell it to ignore it, but on WSS it is not
liking it. We are still looking at the issue and are about to update the
certificate to verify.
Original comment by sdid...@gmail.com
on 9 Feb 2012 at 8:03
Fixed the certificate and things got normal in a hurry. Sorry for posting the
issue as the problem was between my keyboard and chair.
Original comment by sdid...@gmail.com
on 12 Feb 2012 at 4:36
Got it. Thanks.
Original comment by tyoshino@chromium.org
on 22 Feb 2012 at 5:30
We actually had problems again with the last release of the socket protocol.
The issue it turns out was not pywebsockets but the CentOS distribution we were
using. We changed over to Fedora and it worked like a champ with the same
configuration. This is probably better posted in the CentOS user group but in
case anyone else has the problem..... Thanks for your patience!
Original comment by sdid...@gmail.com
on 28 Feb 2012 at 5:50
Original issue reported on code.google.com by
sdid...@gmail.com
on 27 Jan 2012 at 8:25