Mailing-List: contact dev-help@apex.incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@apex.incubator.apache.org
Subject: Apex ASF JIRA Migration
MIME-Version: 1.0
From: Chris Nauroth <cnauroth@hortonworks.com>
To: "dev@apex.incubator.apache.org" <dev@apex.incubator.apache.org>
Thread-Topic: Apex ASF JIRA Migration
Thread-Index: AQHQ644rQEoSUWYxhEaCIGUfSJHq3A==
Date: Thu, 10 Sep 2015 06:01:43 +0000
Message-ID: <D2166AD6.2BAF6%cnauroth@hortonworks.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-ID: <9285DC137635044F896D09CD1E713D23@exch080.serverpod.net>
Content-Transfer-Encoding: quoted-printable

Hello Apex devs,

I'd like to share some new information about the ASF JIRA migration,
tracked in issue INFRA-10145.

https://issues.apache.org/jira/browse/INFRA-10145

First, I'd like to take this opportunity to make sure the whole Apex
community knows that there are multiple means of contacting the Apache
infrastructure engineers.  This is documented here.

http://www.apache.org/dev/infra-contact

My experience has always been that the infrastructure team is responsive,
helpful and friendly.  Please feel free to contact them if necessary.  I
typically start by filing an INFRA issue in JIRA, and then follow up in
the HipChat channel if there is no response in JIRA after a few days.

I checked in at the HipChat channel today about our JIRA migration status.
 Gavin, who is handling INFRA-10145 for us, was not available, but the
other infrastructure engineers pointed out a few challenges.  The version
of JIRA in use hosted at Atlassian is different from the version currently
running in Apache.  In order to do an import, the versions must match
exactly, so our import actually would trigger an upgrade of ASF JIRA.
This would require planned downtime with at least 72 hours notice.  After
that, there would be some challenges with making sure imported data aligns
with best practices used in ASF JIRA, such as using roles instead of
groups.  Considering all of this, the infrastructure team's experience is
that imports take a long time, even in the best case.

The infrastructure team and I discussed a few options for moving ahead.

1. Do not migrate to ASF JIRA.  The infrastructure team noted that there
is nothing in the incubation process that mandates moving off of your
Atlassian hosted instance.  I had not been aware of this.

2. Move to ASF JIRA, but skip the import, and start with a clean slate.

3. Move to ASF JIRA now, but allow the import activity to continue in the
background.  Once we start using ASF JIRA though, imported issues would
not be able to land in the same project key.  They'd likely have to remain
under the old project keys used at the Atlassian instance.

4. Keep waiting for the migration with the understanding that it will take
time to complete.

I'd like to gauge the Apex community's opinion on how to proceed.  My own
opinion is that it would be beneficial to move to ASF JIRA, so I'd prefer
to rule out option 1.  One of the benefits of Apex is its integration with
numerous other Apache projects, and it can be useful to share the same
JIRA instance with those other projects.  For example, if you trace the
root cause of an Apex issue into HDFS, and you want to contact an HDFS
engineer, you can ask me for feedback by entering @cnauroth on a comment.
If you remain in the Atlassian instance, it's not guaranteed that
contributors on other Apache projects will have an account there.  If the
issue is confirmed as an HDFS bug, you can then link your Apex issue to
the corresponding HDFS issue for easy navigation.  I don't believe this
would work as cleanly in the Atlassian instance.

My preference is option 2 for simplicity.  However, this has the side
effect of discarding past history.  Does the Apex community consider the
past history to be important?  Is it important enough to preserve the past
history that you're willing to wait longer for a migration?

Please let me know your thoughts.

--Chris Nauroth