[
https://issues.apache.org/jira/browse/IGNITE-21471?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Roman Puchkovskiy updated IGNITE-21471:
---------------------------------------
Description:
sendMessagesOneChannel and sendMessagesTwoChannels block a Netty thread to do
their assertions. We cannot control which Netty thread does what (actually, we
can, but it's very tricky, non-obvious and fragile). As a result, sometimes
(very rarely, though) the latter test flacks.
It seems that in sendMessagesTwoChannels we can just assert that after sending
2 messages using different channels exactly 2 channels get opened.
sendMessagesOneChannel is only used to test that the approach for blocking the
Netty threads used in sendMessagesTwoChannels works (so it's a test for a
test). If we change the latter to not using blocking, the former can be just
removed.
was:
These tests block a Netty thread to do their assertions. We cannot control
which Netty thread does what (actually, we can, but it's very tricky,
non-obvious and fragile). As a result, sometimes (very rarely, though) the
tests flack.
> Change approach for sendMessagesTwoChannels in DefaultMessagingServiceTest
> --------------------------------------------------------------------------
>
> Key: IGNITE-21471
> URL: https://issues.apache.org/jira/browse/IGNITE-21471
> Project: Ignite
> Issue Type: Bug
> Reporter: Roman Puchkovskiy
> Assignee: Roman Puchkovskiy
> Priority: Major
> Labels: ignite-3
> Fix For: 3.0.0-beta2
>
>
> sendMessagesOneChannel and sendMessagesTwoChannels block a Netty thread to do
> their assertions. We cannot control which Netty thread does what (actually,
> we can, but it's very tricky, non-obvious and fragile). As a result,
> sometimes (very rarely, though) the latter test flacks.
> It seems that in sendMessagesTwoChannels we can just assert that after
> sending 2 messages using different channels exactly 2 channels get opened.
> sendMessagesOneChannel is only used to test that the approach for blocking
> the Netty threads used in sendMessagesTwoChannels works (so it's a test for a
> test). If we change the latter to not using blocking, the former can be just
> removed.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)