[
https://issues.apache.org/jira/browse/IMPALA-10472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17279037#comment-17279037
]
ASF subversion and git services commented on IMPALA-10472:
----------------------------------------------------------
Commit f8ed3f67229c70dbc69501c72bdc146b88b55e4b in impala's branch
refs/heads/master from Alexey Serbin
[ https://gitbox.apache.org/repos/asf?p=impala.git;h=f8ed3f6 ]
IMPALA-10472 flag for Kudu connection negotiation timeout
This patch adds --kudu_client_connection_negotiation_timeout_ms flag
to control client-side connection negotiation timeout in the Kudu
client working as a part of the Impala's BE. Since [1] has been
addressed for Kudu C++ client, it makes sense to provide a control knob
to customize the timeout. That should help to address cases where very
busy cluster nodes hosting Kudu tablet servers aren't fast enough to
negotiate a new connection within the default timeout interval (3 sec),
as mentioned in the description of [1].
[1] https://issues.apache.org/jira/browse/KUDU-2966
Change-Id: I1223187318691da47082608356547f6d78144466
Reviewed-on: http://gerrit.cloudera.org:8080/16705
Reviewed-by: Thomas Tauber-Marshall <[email protected]>
Tested-by: Impala Public Jenkins <[email protected]>
> Add a flag to control Kudu client connection negotiation timeout in backend
> ---------------------------------------------------------------------------
>
> Key: IMPALA-10472
> URL: https://issues.apache.org/jira/browse/IMPALA-10472
> Project: IMPALA
> Issue Type: Improvement
> Components: Backend
> Reporter: Alexey Serbin
> Priority: Major
> Labels: Kudu, client, rpc
>
> Since [KUDU-2966|http://issues.apache.org/jira/browse/KUDU-2966] is
> addressed, it's now possible to control the RPC connection negotiation
> timeout from the Kudu client's side (C++ client). To use the newly
> introduced functionality in Impala's back-end, it's necessary to provide a
> control knob for that.
> This should help to address cases where busy/overloaded cluster nodes hosting
> Kudu tablet servers aren't fast enough to negotiate new connections within
> the default timeout interval (in most cases that's about Kudu server's
> connection negotiation threads being scheduled as needed and getting enough
> CPU time). In practice, it's necessary to customize the corresponding
> setting on the Kudu server side: it's controlled by the
> {{\-\-rpc_negotiation_timeout_ms}} flag for {{kudu-master}} and
> {{kudu-tserver}}.
> The idea is to add a gflag named like
> {{\-\-kudu_client_connection_negotiation_timeout_ms}} with default value of
> 3000. The default value 3000 is to keep the new code backwards-compatible.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
