[
https://issues.apache.org/jira/browse/KUDU-2657?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17621267#comment-17621267
]
Alexey Serbin commented on KUDU-2657:
-------------------------------------
IIUC, this issue has been resolved by
https://github.com/apache/kudu/commit/ccbbfb3006314f2c37f3a40bfec355db9fc90e02.
[~raarora], I guess you can resolve this as FIXED once you gain enough
confidence that the current code provides even distribution of read/scan
requests for the same data ranges when running scan operations in
CLOSEST_REPLICA mode.
> Make sure the standard pseudo-random generator is seeded in C++ kudu client
> ---------------------------------------------------------------------------
>
> Key: KUDU-2657
> URL: https://issues.apache.org/jira/browse/KUDU-2657
> Project: Kudu
> Issue Type: Task
> Components: client
> Reporter: Alexey Serbin
> Assignee: Rahul Arora
> Priority: Major
> Labels: beginner, trivial
>
> This is just a TODO item.
> In our client code we use random() in a few places to select a random server
> among several available ones. In case of large number of clients connecting
> to a Kudu cluster it's essential not all the requests are landed at the same
> server. It would be nice to make sure the random generator is seeded: a quick
> inspection of the code didn't reveal any evidence of that right away.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
