Reconnection stops working

[kmike]

I think it stops working after this change: 6f5514c#L1R321

_sudden_disconnect is raising exception now but on_reconnect is usually called after _sudden_disconnect. Exception pops up and prevents reconnection. This is quite easy to reproduce if redis is executed with 'timeout 1' option.

The traceback (not so useful though):

ERROR:brukva.client:Socket closed on remote end
ERROR:root:Exception in callback <functools.partial object at 0x102966470>
Traceback (most recent call last):
  File "/Users/kmike/svn/tornado/tornado/ioloop.py", line 345, in _run_callback
    callback()
  File "/Users/kmike/svn/tornado/tornado/stack_context.py", line 173, in wrapped
    callback(*args, **kwargs)
  File "/Users/kmike/svn/brukva/brukva/adisp.py", line 111, in _send_result
    self.call(self.g.throw(result))
  File "/Users/kmike/dev/planor/realtime/handlers/pubsub.py", line 27, in change_active_sessions_count
    session_count = yield self.redis.async.hincrby('active_sessions', sessionid, diff)
ConnectionError: Socket closed on remote end

[evilkost]

I think it's fixed here: branch reconnect . Can you confirm for your case?

[kmike]

Hi, thanks for working on this!
I can confirm that reconnections works in my case now.

The recent changes seems to broke the assumption that brukva.client.Client.listen callbacks will receive brukva.client.Message instance as a first parameter (sometimes None is received and it causes AttributeErrors in my code). That's easy to fix on my side though. Is this change intended?

[evilkost]

sometimes None is received and it causes AttributeErrors in my code

This is a bug, i am sure it is related to issue #7
Please check latest version: d47eff4, it should work now.

[kmike]

Yes, I can't reproduce AttributeErrors with the latest version.