reef-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Julia (JIRA)" <j...@apache.org>
Subject [jira] [Resolved] (REEF-1638) Remove master evaluator id from EvaluatorManager when it fails during WaitingForEvaluator state
Date Fri, 14 Oct 2016 21:30:20 GMT

     [ https://issues.apache.org/jira/browse/REEF-1638?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Julia resolved REEF-1638.
-------------------------
    Resolution: Won't Fix

We have made a new assumption, that is if master evaluator fails at any stage, don't recover
the system. 

Ref: REEF-1641 https://issues.apache.org/jira/browse/REEF-1641?filter=12335303


> Remove master evaluator id from EvaluatorManager when it fails during WaitingForEvaluator
state
> -----------------------------------------------------------------------------------------------
>
>                 Key: REEF-1638
>                 URL: https://issues.apache.org/jira/browse/REEF-1638
>             Project: REEF
>          Issue Type: Bug
>            Reporter: Julia
>            Assignee: Julia
>            Priority: Minor
>
> Currently, when master evaluator fails during WaitingForEvaluator state, we tolerant
the failure and request another evalautor for the master. However, the master evaluator id
is not removed in evalautor manager in this case. When a new master evalauotr is registered,
we will get exception because we only allow one master evaluator. This bug needs to be fixed.




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

Mime
View raw message