BIND 10 #349: Busyloop in all python processes caused by nonblock reads of bus

[BIND 10 Development]

#349: Busyloop in all python processes caused by nonblock reads of bus
-------------------------------+--------------------------------------------
      Reporter:  vorner        |        Owner:  jinmei   
          Type:  defect        |       Status:  reviewing
      Priority:  major         |    Milestone:           
     Component:  Unclassified  |   Resolution:           
      Keywords:                |    Sensitive:  0        
Estimatedhours:  0.0           |        Hours:  1        
      Billable:  1             |   Totalhours:  0.25     
      Internal:  0             |  
-------------------------------+--------------------------------------------
Changes (by jinmei):

  * hours:  0.0 => 1.0
  * totalhours:  0.25 => 1.25


Comment:

 Replying to [comment:4 vorner]:
 > I need to get used to the changelogs:
 > {{{
 > [bug]                   Michal Vaner
 > Python processes: they no longer take 100% CPU while idle due to a busy
 loop
 > in reading command session in a nonblocking way. (Trac #349, svn rTBD)
 > }}}
 >
 > I somehow expected that if the parameter is only passed trough to
 already-tested
 > functions, it is tested as well. But more tests probably won't hurt
 anyway.
 >
 I believe in test_check_command_block(, _timeout) of r3017 we should
 confirm the operation actually blocks with no timeout emulated in the
 following case of FakeModuleCCSession.group_recvmsg():
 {{{
         if self._timeout == 0:
             return None, None
 }}}

 If I read it correctly, with the current test code we cannot be sure if it
 really "blocks" forever or it blocks for some period with finite timeout
 and then gets a message.

-- 
Ticket URL: <https://bind10.isc.org/ticket/349#comment:6>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development