[ https://issues.apache.org/jira/browse/ARTEMIS-4509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17786701#comment-17786701 ]
ASF subversion and git services commented on ARTEMIS-4509: ---------------------------------------------------------- Commit 54c1ae630cffea20cfcce42d9d7e5b7834ffa6f0 in activemq-artemis's branch refs/heads/main from Robbie Gemmell [ https://gitbox.apache.org/repos/asf?p=activemq-artemis.git;h=54c1ae630c ] ARTEMIS-4509: fail if LibaioContext check fouls before class, nicer in CI than just logging and failing after anyway > ActiveMQTestBase leak check should identify state clearly without need of logs > ------------------------------------------------------------------------------ > > Key: ARTEMIS-4509 > URL: https://issues.apache.org/jira/browse/ARTEMIS-4509 > Project: ActiveMQ Artemis > Issue Type: Task > Affects Versions: 2.31.2 > Reporter: Robbie Gemmell > Assignee: Robbie Gemmell > Priority: Major > Fix For: 2.32.0 > > > The ActiveMQTestBase class does some leak checking on the LibaioContext > resource usage during the test suite. There is a pre-class check that logs > when it first detects a leak happened before the (typically in a > non-ActiveMQTestBase test class), or fails immediately if a previous class > has detected the leak and failed (after), and a post-class check that fails > if things dont hit 0-usage within 30 seconds. > This is fine for local usage but is a bit of a pain for cases where the > pre-class foul happens in a CI env, as it isnt always so easy to find, get, > and examine the full specific class logs to determine the precise status of > when the issue occurred. It would be more useful if the pre-class foul > detection actually just failed immediately (as will already happen after) and > created an error that explicitly identifies the status from the test fail > without needing to examine the logs at all. -- This message was sent by Atlassian Jira (v8.20.10#820010)