reef-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mariia Mykhailova (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (REEF-1304) Create tests which use .NET Poison to validate our failure scenarios
Date Mon, 25 Apr 2016 23:03:12 GMT

    [ https://issues.apache.org/jira/browse/REEF-1304?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15257209#comment-15257209
] 

Mariia Mykhailova commented on REEF-1304:
-----------------------------------------

Poison can be made deterministic by setting failure probability to 1.0 and period of time
during which failure occurs to 0 (with non-zero delay to failure). I'd like to have a set
of tests which fail in a unified way, and differ only in the moment in which failure is injected.

For exception passing, I think that will be possible after REEF-1286, right? Currently driver
has {{org.apache.reef.exception.EvaluatorException: java.lang.Exception: Exception sent, but
can't be deserialized}}

> Create tests which use .NET Poison to validate our failure scenarios
> --------------------------------------------------------------------
>
>                 Key: REEF-1304
>                 URL: https://issues.apache.org/jira/browse/REEF-1304
>             Project: REEF
>          Issue Type: Test
>          Components: REEF.NET
>            Reporter: Mariia Mykhailova
>            Assignee: Mariia Mykhailova
>              Labels: FT
>   Original Estimate: 168h
>  Remaining Estimate: 168h
>
> There are multiple scenarios of failures in our system. We should use REEF.NET Poison
(delayed fault injection mechanism) to simulate these scenarios in test and to validate the
response of our system to them.
> Here is initial list of tests we need to have.
> || N || Failure injected at || Expected events || Additional verification ||
> | 1 | ON_EVALUATOR_ALLOCATED | FailedEvaluator | |
> | 2 | between ON_EVALUATOR_ALLOCATED and ON_CONTEXT_ACTIVE | FailedEvaluator | |
> | 3 | ON_CONTEXT_ACTIVE | FailedEvaluator + FailedContext | FailedEvaluator should have
FailedContexts filled |
> | 4 | between ON_CONTEXT_ACTIVE and ON_TASK_RUNNING | FailedEvaluator + FailedContext
| FailedEvaluator should have FailedContexts filled |
> | 5 | ON_TASK_RUNNING | FailedEvaluator + FailedContext + FailedTask | FailedEvaluator
should have FailedContexts and FailedTask filled |
> | 6 | between ON_TASK_RUNNING and ON_TASK_COMPLETED | FailedEvaluator + FailedContext
+ FailedTask | FailedEvaluator should have FailedContexts and FailedTask filled |
> | 7 | ON_TASK_COMPLETED | FailedEvaluator + FailedContext | FailedEvaluator should have
FailedContexts filled |
> | 8 | after ON_TASK_COMPLETED | FailedEvaluator ? | race against evaluator completion
|



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message