[
https://issues.apache.org/jira/browse/IGNITE-19605?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17728662#comment-17728662
]
Aleksandr commented on IGNITE-19605:
------------------------------------
This test fail happens when previous test used connect command without paired
disconnect, which leaves the periodic session task running.
When starting the nodes in the next test, it so happens that the request for
the node configuration from the JdbcUrlRegistryImpl or any other periodic task
is received while the Micronaut's context is starting in the RestController for
the second node. In this case the
io.micronaut.core.convert.ConversionService#SHARED is cleared and is not yet
populated.
It's not entirely clear what happens next, but most probably the HTTP server
caches the state where there's no corresponding type converter and this leaves
the node in the broken state.
It doesn't seem possible to properly fix this outside of Micronaut, so there's
just a workaround which adds missing disconnect calls.
> ItDeploymentUnitTest is flaky
> -----------------------------
>
> Key: IGNITE-19605
> URL: https://issues.apache.org/jira/browse/IGNITE-19605
> Project: Ignite
> Issue Type: Bug
> Components: compute
> Reporter: Vadim Pakhnushev
> Assignee: Vadim Pakhnushev
> Priority: Major
> Labels: ignite-3
> Fix For: 3.0.0-beta2
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> Even after the IGNITE-19139 was fixed, the {{ItDeploymentUnitTest}} is still
> flaky, the root cause is still looks like the Micronaut's shared converters
> registry. We should look for a better workaround.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
