[
https://issues.apache.org/jira/browse/SAMZA-1748?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Shanthoosh Venkataraman updated SAMZA-1748:
-------------------------------------------
Description:
In the standalone model, a processor can leave and join the group at any point
in time.
This processor reshuffle is referred to as rebalancing which results in
task(work) redistribution amongst other available, live processors in the group.
Processor rebalancing in existing standalone integration tests(junit tests) is
accomplished through clean shutdown of the processors.
However, in real production scenarios, processor rebalancing is triggered
through unclean shutdown and full garbage collection(GC) of the processors.
As a part of this JIRA, a failure test suite covering these production like
scenarios will be added.
Failure test suite will be built using the zopkio test framework and will cover
the following test cases:
1. Force killing the leader processor of the group.
2. Force killing a single follower in the group.
3. Force killing multiple followers in the group.
4. Force killing the leader and a follower in the group.
5. Suspending and resuming the leader of the group.
Since existing standalone integration tests cover event consumption/production
after a rebalancing phase, these new tests will just test the coordination.
We'll iterate on this initial suite and add tests whenever necessary.
was:
In the standalone model, a processor can leave and join the group at any point
in time.
This processor reshuffle is referred to as rebalancing which results in
task(work) redistribution amongst other available, live processors in the group.
Processor rebalancing in existing standalone integration tests(junit tests) is
accomplished through clean shutdown of the processors.
However, in real production scenarios, processor rebalancing is triggered
through unclean shutdown and full garbage collection(GC) of the processors.
As a part of this JIRA, a failure test suite covering these production like
scenarios will be added.
Failure test suite will be built using the zopkio test framework and will cover
the following test cases:
1. Force killing the leader processor of the group.
2. Force killing a single follower in the group.
3. Force killing multiple followers in the group.
4. Force killing the leader and a follower in the group.
5. Suspending and resuming the leader of the group.
Since existing standalone integration tests cover event consumption/production
after a rebalancing phase, these new tests will just test the coordination.
> Standalone failure tests.
> -------------------------
>
> Key: SAMZA-1748
> URL: https://issues.apache.org/jira/browse/SAMZA-1748
> Project: Samza
> Issue Type: New Feature
> Reporter: Shanthoosh Venkataraman
> Assignee: Shanthoosh Venkataraman
> Priority: Major
>
> In the standalone model, a processor can leave and join the group at any
> point in time.
> This processor reshuffle is referred to as rebalancing which results in
> task(work) redistribution amongst other available, live processors in the
> group.
> Processor rebalancing in existing standalone integration tests(junit tests)
> is accomplished through clean shutdown of the processors.
> However, in real production scenarios, processor rebalancing is triggered
> through unclean shutdown and full garbage collection(GC) of the processors.
> As a part of this JIRA, a failure test suite covering these production like
> scenarios will be added.
> Failure test suite will be built using the zopkio test framework and will
> cover the following test cases:
> 1. Force killing the leader processor of the group.
> 2. Force killing a single follower in the group.
> 3. Force killing multiple followers in the group.
> 4. Force killing the leader and a follower in the group.
> 5. Suspending and resuming the leader of the group.
> Since existing standalone integration tests cover event
> consumption/production after a rebalancing phase, these new tests will just
> test the coordination. We'll iterate on this initial suite and add tests
> whenever necessary.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
