Meng Zhu created MESOS-9358:
-------------------------------
Summary: Test
`SlaveRecoveryTest.AgentReconfigurationWithRunningTask` is flaky.
Key: MESOS-9358
URL: https://issues.apache.org/jira/browse/MESOS-9358
Project: Mesos
Issue Type: Bug
Reporter: Meng Zhu
Assignee: Benno Evers
Attachments: AgentReconfigurationWithRunningTask
The fails with:
{noformat}
../../src/tests/slave_recovery_tests.cpp:4797
Failed to wait 15secs for offers2
{noformat}
It is flaky because after partially accept the first offer, the scheduler will
filter the agent for 5 seconds, and depending on the system load that may last
more than 15 seconds:
{noformat}
I1026 14:32:40.976006 78393344 hierarchical.cpp:2388] Filtered offer with
cpus:2 on agent c8805e35-97ee-44d4-9ab9-e7ea8d703c08-S0 for role * of framework
c8805e35-97ee-44d4-9ab9-e7ea8d703c08-0000
I1026 14:32:40.976065 78393344 hierarchical.cpp:1566] Performed allocation for
1 agents in 197914ns
I1026 14:32:41.981853 79466496 hierarchical.cpp:2388] Filtered offer with
cpus:2 on agent c8805e35-97ee-44d4-9ab9-e7ea8d703c08-S0 for role * of framework
c8805e35-97ee-44d4-9ab9-e7ea8d703c08-0000
I1026 14:32:41.981920 79466496 hierarchical.cpp:1566] Performed allocation for
1 agents in 185853ns
I1026 14:32:42.983603 78393344 hierarchical.cpp:2388] Filtered offer with
cpus:2 on agent c8805e35-97ee-44d4-9ab9-e7ea8d703c08-S0 for role * of framework
c8805e35-97ee-44d4-9ab9-e7ea8d703c08-0000
I1026 14:32:42.983659 78393344 hierarchical.cpp:1566] Performed allocation for
1 agents in 179397ns
I1026 14:32:57.363723 78929920 hierarchical.cpp:2388] Filtered offer with
cpus:2 on agent c8805e35-97ee-44d4-9ab9-e7ea8d703c08-S0 for role * of framework
c8805e35-97ee-44d4-9ab9-e7ea8d703c08-0000
I1026 14:32:57.363788 78929920 hierarchical.cpp:1566] Performed allocation for
1 agents in 466500ns
../../src/tests/slave_recovery_tests.cpp:4797: Failure
Failed to wait 15secs for offers2
{noformat}
I think we can either remove the expectation of the second offer (which I think
is not essential to the test) or manually control the clock.
Full log attached.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
