Mailing-List: contact dev-help@yetus.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@yetus.apache.org
MIME-Version: 1.0
In-Reply-To: <5908A89A.1080504@apache.org>
References: <5908A89A.1080504@apache.org>
From: Sean Busbey <sean.busbey@gmail.com>
Date: Tue, 2 May 2017 10:46:17 -0500
Message-ID: <CAN5cbe4G=06-1WwdPzRyDvcQ+Q9CzwdkRfBJ5NO2ZKp0dapTjw@mail.gmail.com>
Subject: Re: Sporadically failing precommit tasks
To: dev <dev@hbase.apache.org>, dev@yetus.apache.org
Content-Type: text/plain; charset=UTF-8
archived-at: Tue, 02 May 2017 15:46:42 -0000

+dev@yetus in case some other community has run into this.

Yetus is supposed to age out images after a time. Do we know about how
long this has been happening?

Maybe we could do a get-apt update before the install call?

On Tue, May 2, 2017 at 10:41 AM, Josh Elser <elserj@apache.org> wrote:
> I've been running into an issue where the PreCommit job fails when building
> the Yetus Docker image because it attempts to fetch a version of Oracle JDK8
> (8u121) where only a newer version is present (8u131).
>
> I've seen my tasks [1] fail as well someone else's [2], but I've also seen
> some others [3] [4] which don't appear to have issues. I've not been able to
> determine why some builds run and others don't.
>
> It seems related to Docker using a cached image that has a "stale" aptitude
> cache, causing apt-get to try to pull the jdk8u121 package instead of the
> new jdk8u131 package. I'm not really sure how best to address it though.
> Thoughts?
>
> - Josh
>
> [1] https://builds.apache.org/job/PreCommit-HBASE-Build/6647/console
> [2] https://builds.apache.org/job/PreCommit-HBASE-Build/6663/console
> [3] https://builds.apache.org/job/PreCommit-HBASE-Build/6664/console
> [4] https://builds.apache.org/job/PreCommit-HBASE-Build/6665/console


-- 
Sean