reef-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Chung (JIRA)" <>
Subject [jira] [Commented] (REEF-1040) Fix a bug in WatcherTest
Date Thu, 18 Feb 2016 19:24:18 GMT


Andrew Chung commented on REEF-1040:

[~gwkim] Good catch! Although I'm not too fond of the {{isFirstRunningMessage}} though. We
should probably think about what exactly does the {{INIT}} message provide for us. As of now,
it seems that it is set prior *only* to starting the user thread on the Java side. A few questions
that I have are:

1. Should we think of {{INIT}} as {{STARTED}} instead? i.e. does {{INIT}} work as a {{isFirstRunningMessage}}?
2. Should we move {{INIT}} to after we start the thread? What are the consequences here on
a {{SUSPEND}} or {{CLOSE}} message, if the job finishes first on the Evaluator side but the
message has not propagated back to the {{Driver}}?
3. What are some guarantees of messages that we provide from the Evaluator side? In this case,
we always seem to send the {{INIT}} message, but we are not guaranteed to send the {{RUNNING}}

Personally, I am fine with the change here to unblock the 0.14 release, but ideally we may
want to state our intentions clearly for each {{State}} from the Evaluator. Perhaps this is
best left for another JIRA item? What does everyone think?

> Fix a bug in WatcherTest
> ------------------------
>                 Key: REEF-1040
>                 URL:
>             Project: REEF
>          Issue Type: Sub-task
>          Components: REEF-IO
>            Reporter: Geon-Woo Kim
>            Priority: Blocker
> Watcher tests sporadically fails especially in Travis CI. The bug should be fixed.

This message was sent by Atlassian JIRA

View raw message