flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Anton Solovev (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (FLINK-1583) TaskManager reregistration in case of a restart
Date Wed, 21 Dec 2016 13:33:58 GMT

     [ https://issues.apache.org/jira/browse/FLINK-1583?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Anton Solovev updated FLINK-1583:
---------------------------------
    Description: 
Currently, the {{InstanceManager}} identifies {{Instance}}'s based on their {{InstanceConnectionInfo}}.
In case of a restarted {{TaskManager}} which tries to register newly at the {{JobManager}},
the {{InstanceManager}} can mistake this {{TaskManager}} as already registered. This can lead
to a corrupted state.

We should identify {{TaskManager}}s based on some ID to distinguish distinct registration
attempts of a restarted {{TaskManager}}. This will improve the system's stability.

  was:
Currently, the {{InstanceManager}} identifies {{Instance}} s based on their {{InstanceConnectionInfo}}.
In case of a restarted {{TaskManager}} which tries to register newly at the {{JobManager}},
the {{InstanceManager}} can mistake this {{TaskManager}} as already registered. This can lead
to a corrupted state.

We should identify {{TaskManager}}s based on some ID to distinguish distinct registration
attempts of a restarted {{TaskManager}}. This will improve the system's stability.


> TaskManager reregistration in case of a restart
> -----------------------------------------------
>
>                 Key: FLINK-1583
>                 URL: https://issues.apache.org/jira/browse/FLINK-1583
>             Project: Flink
>          Issue Type: Bug
>            Reporter: Till Rohrmann
>
> Currently, the {{InstanceManager}} identifies {{Instance}}'s based on their {{InstanceConnectionInfo}}.
In case of a restarted {{TaskManager}} which tries to register newly at the {{JobManager}},
the {{InstanceManager}} can mistake this {{TaskManager}} as already registered. This can lead
to a corrupted state.
> We should identify {{TaskManager}}s based on some ID to distinguish distinct registration
attempts of a restarted {{TaskManager}}. This will improve the system's stability.



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

Mime
View raw message