[
https://issues.apache.org/jira/browse/QPID-1640?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rajith Attapattu updated QPID-1640:
-----------------------------------
Attachment: QPID_1640.patch
This patch contains improvements made to the existing failover test cases.
It also adds a new test case for produce x messages, consume y (<x) messages,
do failover and then produce p and try to consume (x-y + p) messages.
1.The tests are using sync_publish='all' and sync_ack='true' to weed out
duplicates. Since we control the broker shutdown we can prevent the corner
case where the "last message sent or the last messaged acked could be in doubt".
2. The transacted version of the above test is currently disabled as the
cluster does not do full session resume yet.
3. A another test case to be added is to run the above usecase in a loop by
rotating failover between the two brokers.
4. The x to produce (out of total messages) and y to consume before failover is
determined by a random number generator with a given seed. The seed and the
iterations for the loop could be configured using the system properties defined
in cpp.cluster.testprofile
5. The default seed will generate the following pattern for the test case that
runs in a loop.
Produce 6 consume 4
Produce 3 consume 2
Produce 4 consume 3
Produce 1 consume 1
Produce 4 consume 1
Produce 3 consume 1
Produce 6 consume 6
Produce 3 consume 3
Produce 7 consume 3
Produce 4 consume 1
> Improve existing/add more failover tests
> ----------------------------------------
>
> Key: QPID-1640
> URL: https://issues.apache.org/jira/browse/QPID-1640
> Project: Qpid
> Issue Type: Improvement
> Components: Java Client
> Affects Versions: M5
> Reporter: Rajith Attapattu
> Assignee: Rajith Attapattu
> Fix For: M5
>
> Attachments: QPID_1640.patch
>
>
> It would be good to build a few more tests on top of the current
> FailoverTests.
> One idea is to send/receive a certain amount of messages then failover and
> then send/receive the rest of the messages and verify the order of messages
> and the count.
> Another idea is to failover in loop between two brokers while maintaining the
> same session while doing the above in each iteration.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project: http://qpid.apache.org
Use/Interact: mailto:[email protected]
