Agree with Matt, Kengo and Jun on Ambari mpack. I think it has benefits to Bigtop, again if it can be worked on without reducing efforts on priority items. +1
Ganesh On Tue, Jun 9, 2020 at 2:57 AM Kengo Seki <[email protected]> wrote: > > Now is a great time for making this move as HDP is no longer going to be > managed by a company, so BigTop could replace that role in Ambari. > > This mpack allows BigTop to be installed into Ambari, and allows Ambari > to manage the hadoop packages. Effectively, it would increase the reach of > BigTop to be able to be installed into and managed by Ambari. > > I agree with Matt. Many customers of my company are using HDP via > Ambari, and they are worried about upgrading their stack in the > future. > Providing mpack may be able to help them, though supported components > are very limited for now. > So I'm +1 for this vote (but I personally think the vote was not > necessarily required, as Olaf said). > > > Then I have a question same as Evans for Yuqi, Matt, or others > interested in: are you willing to be the maintainer going forward? :) > > I recommend Masahiro Tanaka, who is an Ambari committer and have been > contributed some patches to Bigtop in the last few months. > I'm going to talk to him about it later. > > Regarding CI-related tasks, my apologies for the late work. I'll > update some Jenkins jobs which don't address new distros yet soon. > > Kengo Seki <[email protected]> > > On Tue, Jun 9, 2020 at 4:31 PM Yuqi Gu <[email protected]> wrote: > > > > Thanks Olaf for expressing the concern on priorities of our Bigtop core > > infrastructure. > > > > The Mpack gives convenience to the Bigtop users to deploy and manage > other > > Bigtop components. > > It also would help Bigtop become an easy-to-use BigData software stack > for > > users as an open source solution. > > > > It is the first step that costs. This discussion may just be a beginning > to > > start to put our eyes on Mpack > > which we may give a low priority. > > > > Definitely we should pay more attention to Bigtop CI, Hadoop3 as Olaf > > listed, > > and I surely would like to help to fix the release-1.5 issues driven by > > Kengo. > > > > > > >> > > *Then I have a question same as Evans for Yuqi, Matt, or others > > interestedin: are you willing to be the maintainer going forward? :)* > > > > I'm also very pleased to be the maintainer to follow up. > > > > BRs, > > Yuqi > > > > > > On Tue, 9 Jun 2020 at 13:02, Jun HE <[email protected]> wrote: > > > > > Thanks Olaf for raising those comments/questions. That really helps us > to > > > think more about how this Bigtop-MPack thing could go further. > > > > > > I think Yuqi is trying to decouple Bigtop-Mpack from Ambari component > > > itself in terms of code/patch/building/packaging, so developer can > easily > > > contribute to it (MPack). The discussion on the PR thread comes to > create a > > > new component just like bigtop-jsvc/bigtop-utils things. So for such > change > > > (adding new component) it might worth discussing and voting in > community > > > besides simply reviewing on some patches. > > > > > > Those items Olaf listed are definitely important to push Bigtop > forward, > > > especiall CI from my point of view, and we will focus on them. While > MPack > > > also gives Bigtop community possibility to extend usage and improve > > > influence as the only open source Big Data SW stack, and there are > > > potential contributors here, maybe we just get the ball rolling. > > > > > > Then I have a question same as Evans for Yuqi, Matt, or others > interested > > > in: are you willing to be the maintainer going forward? :) > > > > > > Regards, > > > > > > Jun > > > > > > > > > Matt Andruff <[email protected]> 于2020年6月8日周一 下午10:32写道： > > > > > > > What problem/case is this additional component trying to solve? > > > > > > > > My opinion is that it's an enable meant tool to make it easy to > use > > > > BigTop by having it's configuration managed by Ambari. Now is a > great > > > time > > > > for making this move as HDP is no longer going to be managed by a > > > company, > > > > so BigTop could replace that role in Ambari. > > > > > > > > This mpack allows BigTop to be installed into Ambari, and allows > > > > Ambari to manage the hadoop packages. Effectively, it would > increase the > > > > reach of BigTop to be able to be installed into and managed by > Ambari. > > > > > > > > I hear that you see a lot of other priorities and they sound like > good > > > > priorities. I'm not able to contribute to those priorities as I > don't > > > know > > > > enough in that area/do not understand what's required. I would > > > contribute > > > > right now to Ambari MPack as it's more in my wheelhouse. > > > > > > > > Hope that helps > > > > > > > > Matt > > > > > > > > On Sun, Jun 7, 2020 at 12:07 PM Olaf Flebbe <[email protected]> wrote: > > > > > > > > > Hi, > > > > > > > > > > Sorry, I am super lost about the implications of that vote. > > > > > > > > > > Why do we need a vote in the first place? > > > > > I am familiar with votes for additional committers, PMC additions > and > > > > > releases. For these votes we have the concept of binding votes by > PMC > > > > > members. > > > > > > > > > > Why should I care about this ? > > > > > Does it have legal implications? > > > > > > > > > > Who will maintain this ? > > > > > I am a bit surprised that we do anything with ambari, since we not > even > > > > > have a maintainer for it > > > > > grep ambari MAINTAINERS.txt -> nothing > > > > > > > > > > What problem/case is this additional component trying to solve? > > > > > If you are asking for a vote I would like to have a more elaborate > > > > > explanation. I looked at the linked ticket: Sorry I am totally lost > > > about > > > > > what is this about. If you would have simply done it, I wouldn’t > have > > > > cared. > > > > > > > > > > As I understand it should somehow enable more contributions. > > > > > Really? MPack is a amabari only technology, that seems not to even > be > > > > > defined properly. How could this ever end in more contributions to > > > > Bigtop ? > > > > > > > > > > What will happen if someone is -1 on this vote ? > > > > > > > > > > If a contributor cares about adopting Bigtop to new business cases, > > > > that’s > > > > > very fine to me. Please go ahead. But please don’t ask for a vote. > All > > > > > committers still have the chance to reject patches. > > > > > > > > > > Generally I am concerned that we care about an (from Bigtop side) > > > > > unmaintained project while neglecting our core infrastructure and > > > > packages > > > > > at the same time. > > > > > 1) The ci.bigtop.apache.org certificate expired again > > > > > 2) The bigtop-trunk-packages do not reflect the architectures we > ought > > > to > > > > > support (ubuntu-18.04 for instance) > > > > > 3) Trunk-packages for amd64 looks like garbage > > > > > 4) Still no-one working on Hadoop3 > > > > > 5) CI needs to be redone (with pipeline jobs) . > > > > > IMO we might have wrong priorities. > > > > > > > > > > Best, > > > > > Olaf > > > > > > > > > > > > > > > > > > > > > > > > > > Am 04.06.2020 um 19:04 schrieb Ganesh Raju < > [email protected]>: > > > > > > > > > > > > +1 > > > > > > > > > > > > On Thu, Jun 4, 2020 at 12:01 PM Matt Andruff < > > > > [email protected]> > > > > > > wrote: > > > > > > > > > > > >> +1 > > > > > >> > > > > > >> On Thu., Jun. 4, 2020, 02:35 Yuqi Gu, <[email protected]> > wrote: > > > > > >> > > > > > >>> Hi folks, > > > > > >>> > > > > > >>> As the discussion with Evans Ye, Jun He, and Matt Andruff (from > > > > Ambari) > > > > > >> on > > > > > >>> > https://github.com/apache/bigtop/pull/555#issuecomment-635228180: > > > > > >>> > > > > > >>> For users / Bigtopers could easily modify Mapck and make > > > > contributions > > > > > to > > > > > >>> it like bugfix and adding more services, > > > > > >>> How about to decouple Mpack from Ambari and put Mpack as a > > > standalone > > > > > >>> component in Bigtop, like > > > *bigtop-packages/src/common/bigtop-mpack*? > > > > > >>> If so, it's convenient for users to get and install Mpack > rpm/deb > > > > > >> packages > > > > > >>> in their own Ambari cluster. > > > > > >>> > > > > > >>> The VOTE: > > > > > >>> [ ] +1, Agree to put Mpack as a standalone component. > > > > > >>> [ ] +0, I don't care either way. > > > > > >>> [ ] -1, do *not * Agree. > > > > > >>> > > > > > >>> Look forward to your comments and feedback, thanks. > > > > > >>> > > > > > >>> BRs, > > > > > >>> Yuqi > > > > > >>> > > > > > >> > > > > > > > > > > > > > > > > > > -- > > > > > > IRC: ganeshraju@#linaro on irc.freenode.ne < > http://irc.freenode.net/ > > > >t > > > > > > > > > > > > > > > > > > -- IRC: ganeshraju@#linaro on irc.freenode.ne <http://irc.freenode.net/>t
