[
https://issues.apache.org/jira/browse/THRIFT-2463?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13967739#comment-13967739
]
Benoit Sigoure commented on THRIFT-2463:
----------------------------------------
I'm wondering whether this could be caused by a spurious SIGALRM timeout,
because {{TestServer.py}} arms the alarm for only 2 seconds and it's possible
that if the machine was under heavy load, it would take more than 2 seconds for
{{TestServer.py}} to start accepting connections.
Excerpt from the code:
{code}
elif server_type == "TProcessPoolServer":
import signal
from thrift.server import TProcessPoolServer
server = TProcessPoolServer.TProcessPoolServer(processor, transport,
tfactory, pfactory)
server.setNumWorkers(5)
def set_alarm():
def clean_shutdown(signum, frame):
print "SIGALRM: Shutting down server"
for worker in server.workers:
if options.verbose > 0:
print 'Terminating worker: %s' % worker
worker.terminate()
if options.verbose > 0:
print 'Requesting server to stop()'
try:
server.stop()
except:
pass
signal.signal(signal.SIGALRM, clean_shutdown)
signal.alarm(2)
set_alarm()
{code}
I'm sure why the alarm is even armed here, and why so aggressively. This was
added in THRIFT-1094 by [~bryanduxbury].
> test/py/RunClientServer.py fails sometimes
> ------------------------------------------
>
> Key: THRIFT-2463
> URL: https://issues.apache.org/jira/browse/THRIFT-2463
> Project: Thrift
> Issue Type: Bug
> Components: Python - Library
> Affects Versions: 0.9.1
> Reporter: Benoit Sigoure
>
> This test is flaky in our CI system and the flakiness is also visible on
> Jenkins (see examples of failed builds
> [here|http://mail-archives.apache.org/mod_mbox/thrift-dev/201309.mbox/%3C1012268498.737.1378727603714.JavaMail.hudson@aegis%3E]
> and [here|https://www.mail-archive.com/[email protected]/msg10124.html]).
> All the test cases of {{AcceleratedBinaryTest}} fail with this signature:
> {code}
> ======================================================================
> ERROR: testByte (__main__.AcceleratedBinaryTest)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File "./TestClient.py", line 85, in setUp
> self.transport.open()
> File
> "./../../lib/py/build/lib.linux-i686-2.7/thrift/transport/TTransport.py",
> line 149, in open
> return self.__trans.open()
> File "./../../lib/py/build/lib.linux-i686-2.7/thrift/transport/TSocket.py",
> line 99, in open
> message=message)
> TTransportException: Could not connect to localhost:9090
> {code}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
