nifi-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joey Echeverria <joe...@gmail.com>
Subject Re: Next release?
Date Tue, 03 Nov 2015 17:05:11 GMT
+1. There's a huge advantage, both from a code quality and community building perspective to
have Committers and Contributors be on a level playing field when it comes to requiring a
review from another Committer. 

-Joey

> On Nov 3, 2015, at 07:27, Alan Jackoway <alanj@cloudera.com> wrote:
> 
> I am not a committer, but I think that at a minimum another committer
> should sign off on it. I don't mind if a different committer says "looks
> good to me, you can merge that," but I don't think committers should put
> their own code in without sign off.
> 
> On Tue, Nov 3, 2015 at 10:23 AM, Oleg Zhurakousky <
> ozhurakousky@hortonworks.com> wrote:
> 
>> May I suggest something that works so well in multitude of projects - one
>> must never merge its own PR, essentially ensuring that there is a consensus
>> 
>> Sent from my iPhone
>> 
>>> On Nov 3, 2015, at 09:00, Joe Witt <joe.witt@gmail.com> wrote:
>>> 
>>> Ricky,
>>> 
>>> Might I remind you, Sir, that you have the power to push!  :-)
>>> 
>>> Let's make sure all the deps are understood (how large?) and that
>>> licensing is fully accounted for.  As long as you have a good plus one
>>> and we're sure its good let's push.  Happy to work with you on it.
>>> 
>>> Also be sure to move the ticket to the 040 release.  Do you have
>>> privileges for that already?
>>> 
>>> Thanks
>>> Joe
>>> 
>>>> On Tue, Nov 3, 2015 at 1:49 PM, Ricky Saltzer <ricky@cloudera.com>
>> wrote:
>>>> Big +1 for these features! I have a pull request out right now for
>> adding a
>>>> Riemann processor <https://github.com/apache/nifi/pull/91>. I've been
>> using
>>>> it on our internal cluster for the past few weeks without any issues,
>> so it
>>>> might be worth taking one last look and then possibly merge in for the
>>>> release on the 19th.
>>>> 
>>>> 
>>>>> On Tue, Nov 3, 2015 at 7:34 AM, Joe Witt <joe.witt@gmail.com> wrote:
>>>>> 
>>>>> Team,
>>>>> 
>>>>> As we work toward an 0.4.0 release here are the current highlights
>>>>> I've captured from the current and resolved tickets.  I might have
>>>>> missed key points but these seem (to me) like the major points:
>>>>> 
>>>>> Version 0.4.0
>>>>> 
>>>>> Highlights of the 0.4.0 release include:
>>>>> - Added proper support for tailing log files.
>>>>> - Updated the framework/UX to support new authentication mechanisms
>>>>> based on username/password
>>>>> - New processor to support Python/Jython scripts as processors.
>>>>> - New processors to capture syslog data received via UDP/TCP
>>>>> - Improved behavior of Execute and Put SQL processors
>>>>> - Provided documentation to help the 'Getting Started' process
>>>>> - Improved efficiency and file handling for merges/sessions dealing
>>>>> with 1000s of objects
>>>>> - New processors to List and Fetch data via SFTP
>>>>> - Improved Kerberos ticket re-registration for HDFS processors
>>>>> - Added processors to interact with Couchbase
>>>>> - Increased convenience when searching for provenance events of a
>>>>> given component
>>>>> - Added SSL support to JMS processors
>>>>> 
>>>>> Now, we have many outstanding tickets still assigned to 0.4.0 which
>>>>> are unresolved.  I reassigned many but still many remain.  Please do
a
>>>>> scan through if you reported them and see which ones can be moved off
>>>>> of 040.
>>>>> 
>>>>> We released 0.3.0 on Sep 19th.  I suggest we try to target Nov 19th
>>>>> then for 0.4.0.  There is already quite a lot in this and so I think
>>>>> we should get very specific about the items remaining which really
>>>>> must be in 040 vs which we can push forward.
>>>>> 
>>>>> I'll keep pairing down the tickets on 040 and pinging folks to
>>>>> understand likely target dates for completion.
>>>>> 
>>>>> Thanks
>>>>> Joe
>>>>> 
>>>>>> On Mon, Nov 2, 2015 at 3:06 PM, Joe Witt <joe.witt@gmail.com>
wrote:
>>>>>> The current process is outlined in our release guide.  But the main
>> idea
>>>>> is
>>>>>> that all who wish to participate in release validation do so from
the
>> RC.
>>>>>> Unit tests are of course run by the builds but we rely on people
>> power to
>>>>>> verify system level testing and that is part of that testing folks
>> should
>>>>>> do.  We obviously can't test all the things and environments and
so on
>>>>> with
>>>>>> this model.  The more CI we can get established the better we can
do.
>>>>> But
>>>>>> we have much room for improvement in validating releases.
>>>>>> 
>>>>>>> On Nov 2, 2015 10:00 AM, "Rick Braddy" <rbraddy@softnas.com>
wrote:
>>>>>>> 
>>>>>>> Joe,
>>>>>>> 
>>>>>>> This reminds me... are there any entry or exit criteria (from
a
>> defects
>>>>>>> perspective) established for NiFi releases?  In other words,
what is
>> the
>>>>>>> criteria for determining when the code is ready for release and
>>>>> production
>>>>>>> use?
>>>>>>> 
>>>>>>> Thanks
>>>>>>> Rick
>>>>>>> 
>>>>>>> -----Original Message-----
>>>>>>> From: Joe Witt [mailto:joe.witt@gmail.com]
>>>>>>> Sent: Monday, November 02, 2015 8:56 AM
>>>>>>> To: dev@nifi.apache.org
>>>>>>> Subject: Re: Next release?
>>>>>>> 
>>>>>>> Team...we def need to address or move a good bit of ticketage
to move
>>>>>>> towards an RC.  It isn't critical we do it 'now' but we should
strive
>>>>> for 6
>>>>>>> to 8 week release cycles in my view.
>>>>>>> 
>>>>>>> We should also decouple the framework/app releases from those
of
>>>>>>> processors in my view but we can kick off another thread for
>> discussion
>>>>>>> there.
>>>>>>> 
>>>>>>> Thanks
>>>>>>> Joe
>>>>>>>> On Oct 29, 2015 11:50 AM, "Joe Witt" <joe.witt@gmail.com>
wrote:
>>>>>>>> 
>>>>>>>> mike - that is good to know.  Look forward to seeing the
ticket.  If
>>>>>>>> you can put the thread dumps up that would obviously be awesome
>> though
>>>>>>>> I recognize why that is non-trivial.
>>>>>>>> 
>>>>>>>> Thanks
>>>>>>>> Joe
>>>>>>>> 
>>>>>>>> On Thu, Oct 29, 2015 at 11:18 AM, Michael Moser <moser.mw@gmail.com
>>> 
>>>>>>>> wrote:
>>>>>>>>> All,
>>>>>>>>> 
>>>>>>>>> On an extremely busy cluster that I work with, I've noticed
some
>>>>>>>>> thread starvation issues on the NCM.  It manifests as
the "spinning
>>>>>>>>> wheel of death" when refreshing the NiFi UI.  Thread
and heap dumps
>>>>>>>>> point to the WebClusterManager in the framework. I've
made some
>>>>>>>>> small quick-win
>>>>>>>> changes
>>>>>>>>> that I'm testing now, but would appreciate feedback from
the
>>>>>>>>> community.
>>>>>>>> I
>>>>>>>>> will write up a ticket shortly that explains it, but
would like to
>>>>>>>>> see it in 0.4.0 if reviewers agree with the changes.
>>>>>>>>> 
>>>>>>>>> Thanks,
>>>>>>>>> -- Mike
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> On Thu, Oct 29, 2015 at 10:04 AM, Joe Witt <joe.witt@gmail.com>
>>>>> wrote:
>>>>>>>>> 
>>>>>>>>>> I haven't done it in a while.  Am happy to take it.
 We need to
>>>>>>>>>> scrub
>>>>>>>> the
>>>>>>>>>> items assigned to 040 and pick our must haves ...
>>>>>>>>>> On Oct 29, 2015 9:20 AM, "Sean Busbey" <busbey@cloudera.com>
>>>>> wrote:
>>>>>>>>>> 
>>>>>>>>>>> Hi Folks!
>>>>>>>>>>> 
>>>>>>>>>>> Tomorrow marks 6 weeks since the 0.3.0 release.
Any one up for
>>>>>>>>>>> starting a release candidate?
>>>>>>>>>>> 
>>>>>>>>>>> --
>>>>>>>>>>> Sean
>>>> 
>>>> 
>>>> 
>>>> --
>>>> Ricky Saltzer
>>>> http://www.cloudera.com
>> 

Mime
View raw message