edgent-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dale LaBossiere (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (EDGENT-445) Console testOverridePortNumber() fails in Jenkins
Date Tue, 30 Jan 2018 20:32:00 GMT

    [ https://issues.apache.org/jira/browse/EDGENT-445?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16345754#comment-16345754
] 

Dale LaBossiere commented on EDGENT-445:
----------------------------------------

Yeah the singleton and test ordering, or not spawning a new JVM for each test class, is the
issue. 

Looks like J8 tests get a new JVM for each test class - i.e., both HttpServerPortTest and
HttpServerTest get new HttpServer instances regardless of which order the tests are run. 
For J7 tests, only the first of those two classes gets a new server instance.  Hence when
HttpServerTest happens first (Jenkins) HttpServerPortTest fails (you can see the actual port
that's reported is the same as the port that got used for the preceding HttpServerTest and
there's no "getInstance" log message indicating a new HttpServer is being created.

> Console testOverridePortNumber() fails in Jenkins
> -------------------------------------------------
>
>                 Key: EDGENT-445
>                 URL: https://issues.apache.org/jira/browse/EDGENT-445
>             Project: Edgent
>          Issue Type: Test
>          Components: Test
>            Reporter: Dale LaBossiere
>            Assignee: Dale LaBossiere
>            Priority: Major
>
> The way the test is currently written doesn't seem to work well in the Jenkins test
environment (the actual port != expected port).  Hmm... maybe something to do with the j8+j7
tests and the "singleton" HttpServer instance?  Would hope that something would fail if for
example the "available port" was no longer available when the server started.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message