Http server appears to be on all the time.

[Neugeniko]

I get this error when clicking Login EVE repeatedly, Login Character out of CREST and trying to Login EVE again and Login EVE after token expires.

 File "/home/neugeniko/src/pyfa/gui/mainFrame.py", line 525, in ssoLogin
   uri = sCrest.startServer()
 File "/home/neugeniko/src/pyfa/service/crest.py", line 102, in startServer
   self.httpd = service.StoppableHTTPServer(('', 6461), service.AuthHandler)
 File "/usr/lib/python2.7/SocketServer.py", line 419, in __init__
   self.server_bind()
 File "/home/neugeniko/src/pyfa/service/server.py", line 60, in server_bind
   BaseHTTPServer.HTTPServer.server_bind(self)
 File "/usr/lib/python2.7/BaseHTTPServer.py", line 108, in server_bind
   SocketServer.TCPServer.server_bind(self)
 File "/usr/lib/python2.7/SocketServer.py", line 430, in server_bind
   self.socket.bind(self.server_address)
 File "/usr/lib/python2.7/socket.py", line 224, in meth
   return getattr(self._sock,name)(*args)
socket.error: [Errno 98] Address already in use 

[blitzmann]

Some more info on our Slack conversation:

Error1 Clicking login eve multiple times and it tries to create another server again on same port and fails. Also fails to launch browser.

Error2 When I logout character I cant login again as it tries to create server on same port and fails.

I do not have these issues with the Windows build. I can click the "Login" button multiple times and apparently start multiple servers. I have a feeling that the server stop command is not functional, as even when the server is supposed to time out after 60 seconds I can still access localhost:6461 which should not happen.

I am using code that I found online for the simple server. Looking more closely at it, the stop command doesn't stop the server itself, but sets a flag so that when a request is made it will honor the request or pass on it. I'll look into fixing this or maybe changing the implementation.

[blitzmann]

Can you please pull from the crest_fitting branch and test? I made a few changes:

• Stop the server when we finish dealing with the returned data from CCP
• Stop the server if it's already running when we try to start it
• Use the socket timeout with handle_timeout instead of threading.Timer which will attempt to close the server if it was already closed.

[Neugeniko]

I still get this error.

Traceback (most recent call last):
  File "/home/neugeniko/src/Pyfa-crest_fitting/gui/crestFittings.py", line 288, in addChar
    uri = sCrest.startServer()
  File "/home/neugeniko/src/Pyfa-crest_fitting/service/crest.py", line 115, in startServer
    self.httpd = service.StoppableHTTPServer(('', 6461), service.AuthHandler)
  File "/usr/lib/python2.7/SocketServer.py", line 419, in __init__
    self.server_bind()
  File "/home/neugeniko/src/Pyfa-crest_fitting/service/server.py", line 60, in server_bind
    BaseHTTPServer.HTTPServer.server_bind(self)
  File "/usr/lib/python2.7/BaseHTTPServer.py", line 108, in server_bind
    SocketServer.TCPServer.server_bind(self)
  File "/usr/lib/python2.7/SocketServer.py", line 430, in server_bind
    self.socket.bind(self.server_address)
  File "/usr/lib/python2.7/socket.py", line 224, in meth
    return getattr(self._sock,name)(*args)
socket.error: [Errno 98] Address already in use

However the server does stop eventually and I'm able to logout and login again. It doesnt crash pyfa.

Same thing in client mode, have to wait between adding characters.

[blitzmann]

Can you please give detailed step by step reproduction steps, and post the debug messages as well? (run pyfa through interpreter with -d). The probably that I'm having is that the server is handled by another thread, so the traceback simply starts with sCrest.startServer() which isn't too helpful.

Also, make sure there's only one pyfa running. The old problematic process may still be running in the background from before the new repo pull since the server goes to a different thread. I did have that probably a couple of times during development

[Neugeniko]

Adding -d switch didnt seem to add anymore information. Steps to reproduce.

1. Load pyfa.
2. Click Login EVE
3. Go thru authentication processes.
4. Click Logout Character
5. Click Login EVE Traceback (most recent call last): File "/home/neugeniko/src/Pyfa-crest_fitting/gui/mainFrame.py", line 525, in ssoLogin uri = sCrest.startServer() File "/home/neugeniko/src/Pyfa-crest_fitting/service/crest.py", line 115, in startServer self.httpd = service.StoppableHTTPServer(('', 6461), service.AuthHandler) File "/usr/lib/python2.7/SocketServer.py", line 419, in init self.server_bind() File "/home/neugeniko/src/Pyfa-crest_fitting/service/server.py", line 60, in server_bind BaseHTTPServer.HTTPServer.server_bind(self) File "/usr/lib/python2.7/BaseHTTPServer.py", line 108, in server_bind SocketServer.TCPServer.server_bind(self) File "/usr/lib/python2.7/SocketServer.py", line 430, in server_bind self.socket.bind(self.server_address) File "/usr/lib/python2.7/socket.py", line 224, in meth return getattr(self._sock,name)(*args) socket.error: [Errno 98] Address already in use
6. Browser fails to launch 7 Closing pyfa fails to terminate the python process.

[blitzmann]

python pyfa.py -d doesn't work? WTF?

I don't really know what to tell you without seing some debug messages and figuring out the logic flow. As it is, it doesn't seem like the server is getting the stop message on your end for unknown reasons.

Keep trying to fiddle with it and collect more data if you can. I'll try to contact you on Slack in the coming days - it's easier to troubleshoot with a real-time platform. But I really want to get whatever this is fixed before we ship on Tuesday. =)

[Neugeniko]

Sorry I did python -d pyfa.py instead of python pyfa.py -d.

Heres the output.

2015-10-29 14:29:34,030 root                     INFO     Starting pyfa
2015-10-29 14:29:44,778 root                     DEBUG    Starting server
2015-10-29 14:30:04,778 service.crest            DEBUG    Handling CREST login with: {'access_token': ['oY6_atX8jwvXGaax-Js-EmiNJ_S8x2U2uakWD2lQo6rRE2VptNVxd6KooFfQ9KZqBpp73taveTGCH3RtlUP2BA2'], 'token_type': ['Bearer'], 'state': ['19f6588d-3608-4caf-b545-6df4cf7e05df'], 'expires_in': ['1200']}
2015-10-29 14:30:04,822 pycrest.eve              DEBUG    Getting resource https://api-sisi.testeveonline.com/
2015-10-29 14:30:04,824 pycrest.eve              DEBUG    Cache miss for resource https://api-sisi.testeveonline.com/ (params={}
2015-10-29 14:30:04,825 pycrest.eve              DEBUG    Getting resource https://api-sisi.testeveonline.com/ (params={})
2015-10-29 14:30:04,867 requests.packages.urllib3.connectionpool INFO     Starting new HTTPS connection (1): api-sisi.testeveonline.com
2015-10-29 14:30:06,377 requests.packages.urllib3.connectionpool DEBUG    "GET / HTTP/1.1" 200 None
2015-10-29 14:30:06,379 pycrest.eve              DEBUG    Getting resource https://sisilogin.testeveonline.com/oauth/verify
2015-10-29 14:30:06,380 pycrest.eve              DEBUG    Cache miss for resource https://sisilogin.testeveonline.com/oauth/verify (params={}
2015-10-29 14:30:06,380 pycrest.eve              DEBUG    Getting resource https://sisilogin.testeveonline.com/oauth/verify (params={})
2015-10-29 14:30:06,382 requests.packages.urllib3.connectionpool INFO     Starting new HTTPS connection (1): sisilogin.testeveonline.com
2015-10-29 14:30:07,868 requests.packages.urllib3.connectionpool DEBUG    "GET /oauth/verify HTTP/1.1" 200 221
2015-10-29 14:30:07,869 service.crest            DEBUG    Got character info: {u'Scopes': u'characterFittingsRead characterFittingsWrite', u'ExpiresOn': u'2015-10-29T03:50:03', u'TokenType': u'Character', u'CharacterName': u'Illbe Frank', u'CharacterOwnerHash': u'QoAn7I4EIAXTYrTgnsuIyn5whf0=', u'CharacterID': 604821304}
2015-10-29 14:30:07,870 root                     DEBUG    Stopping Server
2015-10-29 14:30:13,575 root                     DEBUG    Character logout
2015-10-29 14:30:16,343 root                     DEBUG    Starting server
Traceback (most recent call last):
  File "/home/neugeniko/src/pyfa/gui/mainFrame.py", line 525, in ssoLogin
    uri = sCrest.startServer()
  File "/home/neugeniko/src/pyfa/service/crest.py", line 115, in startServer
    self.httpd = service.StoppableHTTPServer(('', 6461), service.AuthHandler)
  File "/usr/lib/python2.7/SocketServer.py", line 419, in __init__
    self.server_bind()
  File "/home/neugeniko/src/pyfa/service/server.py", line 60, in server_bind
    BaseHTTPServer.HTTPServer.server_bind(self)
  File "/usr/lib/python2.7/BaseHTTPServer.py", line 108, in server_bind
    SocketServer.TCPServer.server_bind(self)
  File "/usr/lib/python2.7/SocketServer.py", line 430, in server_bind
    self.socket.bind(self.server_address)
  File "/usr/lib/python2.7/socket.py", line 224, in meth
    return getattr(self._sock,name)(*args)
socket.error: [Errno 98] Address already in use

[blitzmann]

That is odd, and I'm not sure why it's happening.

I have an idea though. Instead of re initializing the server every time, I might be able to just start and stop handling the requests once it is started. I also need to look into threading a bit more and try to ensure that child threads are killed when the main thread closes.

Also, IIRC you're on debian, correct? I may put together a debian VM to help with debugging

[blitzmann]

@Neugeniko, can you pull latest crest_fitting and try out the newest changes? It works for me under Debian and Windows

I basically did not understand sockets well enough before. My previous change gave the socket a 60 second timeout. The problem with this is that this apparently prevents the socket from closing until the timeout happens or a request is received.

I changed it up a bit to shorten the timeout so that is can actually check if it's still running, and moving the 60-second window to count how many timeouts have happened, and then set run to false if too many timeouts.

[blitzmann]

After speaking with @Neugeniko, I think it's safe to mark this as resolved.