infra-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jacques Le Roux (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (INFRA-15394) Buildbot sends repetitive success, or failure, reports,
Date Mon, 01 Jan 2018 07:24:00 GMT

     [ https://issues.apache.org/jira/browse/INFRA-15394?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Jacques Le Roux updated INFRA-15394:
------------------------------------
    Status: Waiting for Infra  (was: Closed)

Hi Gavin, Taher,

Gavin, I reviewed your change again and we have a problem with it, let me explain.

OFBiz is now 2 separate products in the trunk and the R17.12 branch: the framework and the
plugins supported by 2 sub-branches.

That's why I created the on-ofbiz-framework-commit and the on-ofbiz-plugins-commit schedulers
because we want to test them separately.

Logically:
* The ofbiz-trunk-framework builder runs when a framework change is committed.
* The ofbiz-trunk-plugins builder runs when a plugins change is committed (this has been removed).

But we also want to test the plugins when testing a framework only change, in case it has
an impact on the plugins tests. That's why we have also the ofbiz-trunk-framework-plugins
builder in the on-ofbiz-framework-commit scheduler.

Also there are no ways to independently test the plugins, they must be tested with the framework
(they can't be used independently). Hence the on-ofbiz-plugins-commit scheduler which runs
all the tests (framework and plugins).

So it seems there is a contradiction:
# we want to test them separately
# but we also want to test the plugins when changing only the framework.

And I don't know Buildbot enough to do it in another way I did.

So, currently, changes in the plugins are only tested when a change in the framework is done.
It could be seens as acceptable, and we would wait a change in the framework to tests the
plugins. But this could lead to have several commits done in plugins before we detect an issue
in those commits. And that can complicate things when identifying the culprit commit/s, the
sooner the better.

So with these specifications maybe you know a better way to achieve that than the way I did?

Notes:
* there is yet no builder for the R17 branch but it should be quickly done based on the trunk
builders.
* The ofbiz-trunk-plugins-rat builder, currently commented out, is useless and should be removed.

> Buildbot sends repetitive success, or failure, reports, 
> --------------------------------------------------------
>
>                 Key: INFRA-15394
>                 URL: https://issues.apache.org/jira/browse/INFRA-15394
>             Project: Infrastructure
>          Issue Type: Bug
>          Components: Buildbot
>            Reporter: Jacques Le Roux
>            Assignee: Gavin
>             Fix For: Dec 2017
>
>
> For some months, Buildbot sends repetitive success, or failure, reports, when it should
be only triggered on state changes. It's not a big deal but it would be good if we could find
the reason.
> It started around mid-June it seems (19 June more exactly). Could that be due to our
script? I can see no reasons within it.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message