[jira] [Closed] (PROTON-1851) [python] Unable to send messages to newly connected server after reconnect has happened

["Alan Conway (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/PROTON-1851?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Alan Conway closed PROTON-1851.
-------------------------------
    Resolution: Not A Problem
      Assignee: Alan Conway

The test is incorrect, because stopping the python reactor does not actually close any of its connections. Since server1 is in the same process as the test, it's connections don't get closed at all - the server just stops responding, but there's no socket close to trigger a re-connect.

There's no easy way with the current python reactor to simulate a connection failure within a single process. It may become easier after [~astitcher] does his planned re-work of the python IO layer.

The other solution is to build a multi-process test, which is probably a more accurate in this case anyway (actually you really need a multi-host test with a network that can be disabled to fully test failover but that's not in scope for our basic repo testing) There are muti-process tests in ruby/tests, cpp/tests, c/tests and go/tests - the c and cpp tests use python as the test driver so that could be adapted.

A side note on print statements in the tests: you can use the environment variables PN_TRACE_FRM=1 PN_TRACE_LOG=1 or PN_TRACE_EVT=1 to get a lot of useful info from the underlying C library without modifying the test code. In python you can also enable logging from the python binding code with

    logging.basicConfig(level=logging.DEBUG)

at the start of the script.

 

 

 

 

 

 

> [python] Unable to send messages to newly connected server after reconnect has happened
> ---------------------------------------------------------------------------------------
>
>                 Key: PROTON-1851
>                 URL: https://issues.apache.org/jira/browse/PROTON-1851
>             Project: Qpid Proton
>          Issue Type: Bug
>          Components: python-binding
>    Affects Versions: proton-c-0.22.0
>            Reporter: Jiri Daněk
>            Assignee: Alan Conway
>            Priority: Major
>
> If I specify multiple {{container.connect(urls=[...]}}, my client does reconnect fine in case of server failure. What does not seem to work for me is making the client send messages to the new server. I think PROTON-1515 may be related to my issues.
> h4. Test case
> Can be run as {{python -m unittest proton_tests.engine.ServerTest.testFailover}} in {{qpid-proton/tests/python}}
> {code:title=tests/python/proton_tests/engine.py}
> class ServerTest(Test):
>   def testFailover(self):
>     """ Verify that messages continue to be delivered
>     (to the new broker) after failover happens"""
>     server1 = common.TestServer2()
>     server1.start()
>     server2 = common.TestServer2()
>     server2.start()
>     print("testFailover")
>     class Program(MessagingHandler):
>       first = True
>       def on_start(self, event):
>         print("sender: on start")
>         # self.conn = event.container.connect(url="%s:%s" % (server1.host, server1.port))  # , allowed_mechs="ANONYMOUS")
>         self.conn = event.container.connect(urls=["%s:%s" % (server1.host, server1.port), "%s:%s" % (server2.host, server2.port)])
>       def on_sendable(self, event):
>         print("sender: on sendable")
>         message = Message()
>         self.sender.send(message)
>       def on_connection_opened(self, event):
>         self.sender = event.container.create_sender(self.conn, "some_address")
>         print("sender: on connection opened")
>     p = Program()
>     c = Container(p)
>     t = Thread(target=c.run)
>     t.start()
>     sleep(1)
>     server1.stop()
>     print("server 1 stopped")
>     sleep(1)
>     server2.stop()
>     sleep(1)
>     c.stop()
>     print("msgs: ", len(server1.tags), len(server2.tags))
>     assert len(server1.tags) > 0
>     assert len(server2.tags) > 0  # this assert fails, len == 0
> {code}
> {code:title=tests/python/proton_tests/common.py}
> class TestServer2(MessagingHandler):
>   """ Base class for creating test-specific message servers.
>   """
>   def __init__(self, **kwargs):
>     super(TestServer2, self).__init__()
>     self.args = kwargs
>     self.reactor = Container(self)
>     self.host = "127.0.0.1"
>     self.port = 0
>     if "host" in kwargs:
>       self.host = kwargs["host"]
>     if "port" in kwargs:
>       self.port = kwargs["port"]
>     self.handlers = [CFlowController(10), CHandshaker()]
>     self.thread = Thread(name="server-thread", target=self.run)
>     self.thread.daemon = True
>     self.running = True
>     self.tags = []
>   def start(self):
>     self.thread.start()
>   def stop(self):
>     self.running = False
>     self.reactor.wakeup()
>     self.acceptor.close()
>     self.reactor.stop()
>     self.thread.join()
>   # Note: all following methods all run under the thread:
>   def run(self):
>     self.reactor.run()
>   def on_start(self, event):
>       print("on start server")
>       retry = 0
>       if self.port == 0:
>           self.port = str(randint(49152, 65535))
>           retry = 10
>       while retry > 0:
>           try:
>               # self.acceptor = self.reactor.acceptor(self.host, self.port)
>               self.acceptor = event.container.listen(Url("%s:%s" % (self.host, self.port)))
>               break
>           except IOError as e:
>               print(e)
>               self.port = str(randint(49152, 65535))
>               retry -= 1
>       assert retry > 0, "No free port for server to listen on!"
>   def on_delivery(self, event):
>     """
>     :type event: proton.Event
>     """
>     print("on delivery")
>     event.delivery.settle()
>     self.tags.append(event.delivery.tag)
> {code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@qpid.apache.org
For additional commands, e-mail: dev-help@qpid.apache.org