[
https://issues.apache.org/jira/browse/ZOOKEEPER-2152?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15282248#comment-15282248
]
Michael Han commented on ZOOKEEPER-2152:
----------------------------------------
There are a set of failures in TestReconfig.cc that can be reproduced locally
(Ubuntu 14.04) against trunk. I ran TestReconfig 100 times, in test_mt and
test_st mode separately. test_st is more stable (2 failure out of 100), and
test_mt has more failures (17 failure out of 100). The set of failures are:
{code:title=TestReconfig.cc:154: Assertion: assertion failed [Expression:
false] (15 failures out of 100 in test_mt) |borderStyle=solid}
// Else we've looped around!
else if (first == next)
{
CPPUNIT_ASSERT(false);
}
{code}
{code:title=TestReconfig.cc:474: Assertion: assertion failed [Expression: found
!= string::npos, 10.10.10.4:2004 not in newComing list]
(1 failure out of 100 in test_mt) |borderStyle=solid}
// Assert next server is in the 'new' list
size_t found = newComing.find(next);
CPPUNIT_ASSERT_MESSAGE(next + " not in newComing list", found !=
string::npos);
{code}
{code:title=TestReconfig.cc:183: Assertion: equality assertion failed
[Expected: 1, Actual : 0]
(1 failure out of 100 in test_mt) |borderStyle=solid}
void setServersAndVerifyReconfig(const string new_hosts, bool is_reconfig)
{
setServers(new_hosts);
CPPUNIT_ASSERT_EQUAL(is_reconfig, isReconfig());
}
{code}
{code:title=TestReconfig.cc:381: Assertion: assertion failed [Expression:
numClientsPerHost.at(index) >= lowerboundClientsPerServer(numClients,
numServers)]
(1 failure out of 100 in test_st) |borderStyle=solid}
CPPUNIT_ASSERT(numClientsPerHost.at(index) >=
lowerboundClientsPerServer(numClients, numServers));
numClientsPerHost.at(index) = 0; // prepare for next test
{code}
{code:title=TestReconfig.cc:573: Assertion: assertion failed [Expression:
numClientsPerHost.at(i) >= lowerboundClientsPerServer(numClients, numServers)]
(1 failure out of 100 in test_st) |borderStyle=solid}
for (int i = 0; i < numServers; i++) {
CPPUNIT_ASSERT(numClientsPerHost.at(i) <=
upperboundClientsPerServer(numClients, numServers));
CPPUNIT_ASSERT(numClientsPerHost.at(i) >=
lowerboundClientsPerServer(numClients, numServers));
numClientsPerHost.at(i) = 0; // prepare for next test
}
{code}
The failures are align with what [~suda] has observed, though I have a much
lower reproduce rate (17%) comparing to 50%. In particular, the failure
described in the JIRA has a reproduce rate of 1%. Based on the results, I think
the reconfig c client tests are more fragile in threaded mode. Will investigate
further on the states of object when the tests failed.
> Intermittent failure in TestReconfig.cc
> ---------------------------------------
>
> Key: ZOOKEEPER-2152
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2152
> Project: ZooKeeper
> Issue Type: Sub-task
> Components: c client
> Reporter: Michi Mutsuzaki
> Assignee: Michael Han
> Labels: reconfiguration
> Fix For: 3.6.0
>
>
> I'm seeing this failure in the c client test once in a while:
> {noformat}
> [exec]
> /home/jenkins/jenkins-slave/workspace/ZooKeeper-trunk/trunk/src/c/tests/TestReconfig.cc:474:
> Assertion: assertion failed [Expression: found != string::npos,
> 10.10.10.4:2004 not in newComing list]
> {noformat}
> https://builds.apache.org/job/ZooKeeper-trunk/2640/console
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
