BIND 10 #2410: race condition(?) in dhcp[46]_test.py

[BIND 10 Development]

#2410: race condition(?) in dhcp[46]_test.py
-------------------------------------+-------------------------------------
                   Reporter:  jelte  |                 Owner:  stephen
                       Type:         |                Status:  accepted
  defect                             |             Milestone:  Sprint-
                   Priority:         |  DHCP-20121129
  medium                             |            Resolution:
                  Component:  dhcp   |             Sensitive:  0
                   Keywords:         |           Sub-Project:  DHCP
            Defect Severity:  N/A    |  Estimated Difficulty:  4
Feature Depending on Ticket:         |           Total Hours:  0
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------

Comment (by stephen):

 Now ready for review.

 The problem appears to have been diagnosed correctly - the parent process
 reads from the pipe connecting it to the subprocess, assuming that at that
 point, the subprocess had written to the pipe.  Although not encountering
 the problem on my systems, I was able to simulate it by (temporarily)
 adding a 2-second delay into "main()" of the server process.

 This fix causes the parent to loop round reading the pipe until either the
 parent process terminates or some output appears.

 Regarding the unclosed file, additional close calls were added to close
 unclosed descriptors, and the !ProcessInfo object explicitly deleted.  The
 latter appears to eliminate all but one of the unclosed descriptor
 messages, the exception being the case where the process ends by itself
 (instead of being killed).  It is not clear what the reason for this is.

-- 
Ticket URL: <http://bind10.isc.org/ticket/2410#comment:6>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development