[
https://issues.apache.org/jira/browse/MESOS-9083?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16713230#comment-16713230
]
Meng Zhu commented on MESOS-9083:
---------------------------------
This is caused by MESOS-9460.
In the bad log
- the first reservation is processed on both the master and agent
- the unreserve operation is processed on the master (including the allocator)
- agent sends the updateSlaveMessage to master which overwrites the unreserve
operation
- the unreserve operation is processed on the agent
Now the resource view on the master and agent is out-of-sync.
> Test ReservationEndpointsTest.ReserveAndUnreserveNoAuthentication is flaky.
> ---------------------------------------------------------------------------
>
> Key: MESOS-9083
> URL: https://issues.apache.org/jira/browse/MESOS-9083
> Project: Mesos
> Issue Type: Bug
> Components: allocation
> Reporter: Meng Zhu
> Assignee: Meng Zhu
> Priority: Major
> Labels: flaky, mesosphere
> Attachments:
> ReservationEndpointsTest_ReserveAndUnreserveNoAuthentication_badrun.txt
>
>
> Flaky on CI debian-9-SSL_GRPC
> Error Message
> {noformat}
> ../../src/tests/reservation_endpoints_tests.cpp:1430
> Value of: (response)->status
> Actual: "409 Conflict"
> Expected: Accepted().status
> Which is: "202 Accepted"
> Body: "ports:[31000-32000]; cpus(reservations: [(DYNAMIC,role)]):1;
> disk:35132; mem(reservations: [(DYNAMIC,role)]):512 does not contain cpus:1"
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
