You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kudu.apache.org by "Alexey Serbin (JIRA)" <ji...@apache.org> on 2017/05/24 02:00:08 UTC

[jira] [Updated] (KUDU-2021) Add an integration test: in case of multi-master, RPC is retried in case of negotiation error with master

     [ https://issues.apache.org/jira/browse/KUDU-2021?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Alexey Serbin updated KUDU-2021:
--------------------------------
    Description: 
In case of multi-master Kudu cluster the client is supposed to retry a failed RPC with other master server.

For example, when the leader master is up but "frozen", if the client times out on negotiating a connection, the RPC fails and should be retried with other available masters.

This is about adding an integration test to cover functionality similar to KUDU-1580, but master-related.

  was:
In case of multi-master Kudu cluster the client is supposed to retry a failed RPC with other master server, but that not the case if RPC fails on connection negotiation.

For example, when the leader master is up but "frozen", if the client times out on negotiating a connection, the RPC fails and is not retries with other available masters.

This is similar to KUDU-1580, but for masters.

     Issue Type: Task  (was: Bug)
        Summary: Add an integration test: in case of multi-master, RPC is retried in case of negotiation error with master  (was: In case of multi-master, connection negotiation error to leader master is not retried)

> Add an integration test: in case of multi-master, RPC is retried in case of negotiation error with master
> ---------------------------------------------------------------------------------------------------------
>
>                 Key: KUDU-2021
>                 URL: https://issues.apache.org/jira/browse/KUDU-2021
>             Project: Kudu
>          Issue Type: Task
>          Components: client
>    Affects Versions: 1.3.1
>            Reporter: Alexey Serbin
>            Assignee: Alexey Serbin
>              Labels: client, master
>
> In case of multi-master Kudu cluster the client is supposed to retry a failed RPC with other master server.
> For example, when the leader master is up but "frozen", if the client times out on negotiating a connection, the RPC fails and should be retried with other available masters.
> This is about adding an integration test to cover functionality similar to KUDU-1580, but master-related.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)