[
https://issues.apache.org/jira/browse/IGNITE-16375?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17483608#comment-17483608
]
Sergey Korotkov commented on IGNITE-16375:
------------------------------------------
[~av]
Sorry, i don't quite follow what should be done for the
{*}_startClientGrid(cfg)_{*}.
It gets the configuration as a parameter, after that modifies it switching to
the client mode and start the client grid using this config. Most of the tests
pass the default test configuration (which is server one by default) and expect
to hava a client grid started with just the same config as a server one it
start a second before. So if we would raise exception if the incoming config is
server one we would do that for virtually all tests.
Are you 100% sure we need to change the behaviour of the test framework on such
low basic level?
Shouldn't we just force client mode in the {*}_startClientGrid(name)_{*}?
> GridAbstractTest.startClientGrid() method does not guarantee that started
> node will be the client
> -------------------------------------------------------------------------------------------------
>
> Key: IGNITE-16375
> URL: https://issues.apache.org/jira/browse/IGNITE-16375
> Project: Ignite
> Issue Type: Bug
> Reporter: Anton Vinogradov
> Assignee: Sergey Korotkov
> Priority: Major
> Labels: newbie
> Attachments: Client_node_start_failed_reproducer.patch
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> If {{cfg.setClientMode(false)}} was applied to the Ignite configuration,
> {{GridAbstractTest.startClientGrid()}} will start the server node, while
> client node is expected according to the method name and JavaDoc.
> {noformat}
> Starts new client grid with given index.
> {noformat}
> {{GridAbstractTest.startClientGrid()}} should start client node even when
> {{cfg.setClientMode(false)}} is set or throw an exception (preferred).
> See reproducer at [^Client_node_start_failed_reproducer.patch]
--
This message was sent by Atlassian Jira
(v8.20.1#820001)