[Pkg-bazaar-maint] Bug#736158: random FTBFS: test_graceful_shutdown_waits_for_clients_to_stop

[Adam Borowski]

Package: src:bzr
Version: 2.6.0-3
Severity: serious
Justification: fails to build from source (but built successfully in the past)


I'm afraid that bzr randomly (7/8 tries) FTBFSes due to a failure of one
test on speedier machines:

======================================================================
FAIL: bzrlib.tests.test_smart_transport.TestSmartTCPServer.test_graceful_shutdown_waits_for_clients_to_stop
----------------------------------------------------------------------
_StringException: log: {{{
INFO  Requested to stop gracefully
305.285  Stopping SmartServerSocketStreamMedium(client=('127.0.0.1', 60256))
    INFO  Waiting for 1 client(s) to finish
    INFO  Requested to stop gracefully
}}}

Traceback (most recent call last):
  File "/tmp/buildd/bzr-2.6.0/build/lib.linux-x86_64-2.7/bzrlib/tests/test_smart_transport.py", line 1458, in test_graceful_shutdown_waits_for_clients_to_stop
    self.assertFalse(server._fully_stopped.isSet())
  File "/usr/lib/python2.7/unittest/case.py", line 418, in assertFalse
    raise self.failureException(msg)
AssertionError: True is not false

----------------------------------------------------------------------

I repeated the build on an armhf box, all 3 tries succeeded.  So did it on
all buildds for first-class architectures, while on the x32 buildd there
were two failures out of two attempts.  It seems that the x32 buildd runs on
a good deal faster machine than i386 and amd64 ones: looking at some random
package, I see 25s:x32 vs 1m50s:i386.  All other architectures are,
unsuprisingly, slower.

So it's some timing issue.  I have a strong hunch it's a bug in the test
suite rather than in the code being tested, so if it's tricky to debug,
disabling that test wouldn't be as bad an idea as papering over test
failures usually is.

(Not sure what's the right severity for "FTBFSes on too fast".  As my home
machines are not so hot, I went with "serious" as hardware only goes faster,
but you probably know better.)


Meow!
-- System Information:
Debian Release: jessie/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (150, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 3.13.0-rc8-x32+ (SMP w/6 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash