failure in allmydata.test.test_runner.RunNode.test_client: 'does not look like a running node directory'

[davidsarah]

Reported by Drew Perttula:

Here is my one test failure. Version is http://tahoe-
lafs.org/source/tahoe/trunk-hashedformat from an hour ago

(i.e. source:trunk@4536)

[FAIL]: allmydata.test.test_runner.RunNode.test_client

Traceback (most recent call last):
  File "/my/dl/modified/tahoe-new/src/allmydata/test/test_runner.py",
line 504, in _cb4
    self.failUnlessEqual(rc_or_sig, 0, errstr)
twisted.trial.unittest.FailTest: rc=2, OUT: '', ERR: '/my/dl/modified/
tahoe-new/_trial_temp/test_runner/RunNode/test_client/c1 does not look
like a running node directory (no twistd.pid)
'
not equal:
a = 2
b = 0 

Not sure which platform this is.

[zooko]

reported by Drew Perttula:

Not sure which platform this is.

That's interesting! Could you paste in the output from tahoe --version?

[drewp]

% bin/tahoe --version
/usr/lib/python2.6/dist-packages/Pyrex/Compiler/Errors.py:17: 
DeprecationWarning: BaseException.message has been deprecated as of Python 2.6
  self.message = message
allmydata-tahoe: 1.7.0-r4536, foolscap: 0.5.1, pycryptopp: 0.5.19, zfec: 1.4.7, 
Twisted: 8.2.0, Nevow: 0.9.33-r17222, zope.interface: 3.5.1, python: 2.6.2, 
platform: Linux-Ubuntu_9.04-i686-32bit_ELF, sqlite: 3.6.21, simplejson: 2.0.9, 
argparse: 1.1, pycrypto: 2.0.1, pyOpenSSL: 0.7, pyutil: 1.7.9, zbase32: 1.1.2, 
setuptools: 0.6c15dev, pyasn1: 0.0.11a, pysqlite: 2.4.1

[zooko]

drewp: is this reproducible at all? I can't imagine what caused it, and I'm tempted to close it as "cannot reproduce". I suppose it is possible that it was a transient failure in the filesystem or kernel, etc.

[davidsarah]

The test is racy. It even has a comment saying so, just before the _cb4 callback that failed:

# TODO: On a slow machine, the node might kill
# itself before we get a chance too, especially if spawning the
# 'tahoe stop' command takes a while.

drewp: is it possible that the machine was bogged down at the time of this failure?