Yesterday I decided to completely refresh my qpid build area for
qpid, qpid-proton, and qpid-dispatch. For each:
- select master/trunk branch; git update to latest
- git clean -dfx - to get rid of any stale artifacts
- rm -rf /opt/local - fresh install area
- md build; cd build; - cmake for debug, install prefix = /opt/local
- make install
- make test
I observed bunches of intermittent connection errors in qpid and
qpid-dispatch. Each test run yielded a different set of failures.
For instance in dispatch I could run
ctest -VV -R one_router
ten times in a row no problem. Then on another ten runs I'd get three errors
{noformat}
15: FAIL: test_16_multicast_unsettled (system_tests_one_router.RouterTest)
15: ----------------------------------------------------------------------
15: Traceback (most recent call last):
15: File "/home/git/qpid-dispatch/tests/system_tests_one_router.py", line
1066, in test_16_multicast_unsettled
15: self.assertEqual(None, test.error)
15: AssertionError: None != 'Timeout Expired: sent=0, received=0, accepted=0'
{noformat}
The errors are not always in test_16 but are scattered around. After I started
looking at them more closely they were all Timeout Expired and usually
related to SSL or SASL.
Eventually I came across https://issues.apache.org/jira/browse/DISPATCH-340
which was a hard error and completely reproducible. The issue here is that
some dispatch connection query result is expecting a python list to contain
info [A, B, C]. On my system the query result is coming back as [B, A, C].
I fixed the hard error by searching the result list for info item A and
not expecting it to be in slot [0].
So why does every other user and CI system pass this test and mine [See note
1] fails? My system used to pass all the time.
There has been plenty of change to Dispatch and Proton wrt connections lately.
Given that Dispatch-340 is showing that connections are now added in a
different order than the historic pattern then this is where the issue is
lurking. I put some effort into chasing down what's going on but would like some
collective help and suggestions.
-Chuck
[1] Lenovo Thinkpad W541, 16Gb ram, Intel Core i74810MQ CPU @ 2.80GHz x 8,
OS 64-bit, Graphics Intel Haswell Mobile, Gnome V3.18.2, Disk 500GB SSD.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
