Yes, Thank you. If you got any problems, I'd glad to help. I think it's time to 
make it opensource ASAP. 




------------------ ???????? ------------------
??????: "Yang Bo"<oaky...@gmail.com>;
????????: 2018??11??28??(??????) ????11:55
??????: "dev"<dev@servicecomb.apache.org>;

????: Re: [Discuss]Adding Etcd as a config center for Java-Chassis



@bismy

Do you mean the config-center opensource? Yes I can do that.

On Wed, Nov 28, 2018 at 11:07 AM bismy <bi...@qq.com> wrote:

> @Yang Bo
>
>
> Can you start working on service center opensource? I think it's time for
> this task. And there maybe some tasks need to do, like code clean up,
> licences and others staffs. I can find one to help you with these works if
> needed.
>
>
> I agree with you that servicecomb now seems to be a "lackluster system",
> but we are in a quite progressive way to build more, and we are creating
> great components. I am preparing a glossary what we are going to build.
>
>
>
>
> ------------------ ???????? ------------------
> ??????: "Xiaoliang Tian"<xiaoliang.t...@gmail.com>;
> ????????: 2018??11??28??(??????) ????10:45
> ??????: "dev"<dev@servicecomb.apache.org>;
>
> ????: Re: [Discuss]Adding Etcd as a config center for Java-Chassis
>
>
>
> etcd used by service center is not allowed to expose directly to any
> client, this etcd only exposed service center. unless you setup isolated
> etcd cluster for config management
>
> Yang Bo <oaky...@gmail.com> ??2018??11??28?????? ????10:08??????
>
> > @Sure
> > Yes we can use consul. I choose etcd just because service-center already
> > uses it. Actually, if we go with consul, we can even drop service-center,
> > as consul provides service registry/discovery and configuration store. We
> > can keep the schema data as configuration.
> >
> > I've been suggesting to opensource the CC and it's frontend since
> May/June.
> > But here we are, there are still no timeline when this will be done.
> > And without this 2 component, the opensource version(servicecomb) feels
> > like a lackluster system.
> >
> > On Wed, Nov 28, 2018 at 9:11 AM Willem Jiang <willem.ji...@gmail.com>
> > wrote:
> >
> > > If we have plan to open source the config center and governance,
> > > it's could be better we start a discussion here to make a road map.
> > >
> > > Willem Jiang
> > >
> > > Twitter: willemjiang
> > > Weibo: ????willem
> > >
> > > On Wed, Nov 28, 2018 at 8:34 AM wjm wjm <zzz...@gmail.com> wrote:
> > > >
> > > > it's better to make huawei config center opensource
> > > > after governance opensource, can work with huawei config center
> > directly
> > > >
> > > > Yang Bo <oaky...@gmail.com> ??2018??11??27?????? ????2:21??????
> > > >
> > > > > Etcd in itself is good enough to be used as a config center. I
> don't
> > > see
> > > > > the need for an extra layer, it reduces performance and adds
> > > complexity.
> > > > >
> > > > > For example, if we use Etcd directly, the watch mechanism is
> > naturally
> > > > > supported by grpc which uses http2, there's no need for using
> > websocket
> > > > > which might have problems under some firewall rules.
> > > > >
> > > > > I checked the CC(CSE) client implementation in java-chassis, it
> uses
> > > vert.x
> > > > > and http to talk to the server and use websocket for watching when
> > > > > possible.
> > > > >
> > > > > It have implemented tenant, environment separation and we can
> achieve
> > > this
> > > > > by using proper key prefixes in Etcd(v3). I don't see a particular
> > > feature
> > > > > impossible to implement using Etcd only.
> > > > >
> > > > > Besides, it takes a lot of time and effort to opensource the
> > > Config-Center.
> > > > >
> > > > > There is one minor issue for Etcd though. The official Etcd client
> > for
> > > java
> > > > > jetcd is not published to central maven repository yet. They are
> > > preparing
> > > > > an official 0.3.0 release but does not have a timeline for it.
> > > > >
> > > > > On Tue, Nov 27, 2018 at 9:50 AM bismy <bi...@qq.com> wrote:
> > > > >
> > > > > > Do you have any evaluation document by using etcd as config
> center?
> > > Now
> > > > > > config-cc supports using Huawei Config Center as the backend. And
> > > this
> > > > > > implementation is based on ectd with an extra abstraction like
> > > service
> > > > > > center. I am not quite sure if using etcd itself can fitful for
> > > users use
> > > > > > cases and if it is a good solution or just a workable solution.
> > > > > >
> > > > > >
> > > > > > Maybe we can make Huawei Config Center opensource and distribute
> it
> > > like
> > > > > > service center which is quite simple too.
> > > > > >
> > > > > >
> > > > > > ------------------ ???????? ------------------
> > > > > > ??????: "Yang Bo"<oaky...@gmail.com>;
> > > > > > ????????: 2018??11??26??(??????) ????7:30
> > > > > > ??????: "dev"<dev@servicecomb.apache.org>;
> > > > > >
> > > > > > ????: [Discuss]Adding Etcd as a config center for Java-Chassis
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi,
> > > > > >
> > > > > > The ServiceComb project does not have a usable config-center and
> > many
> > > > > users
> > > > > > have shown concerns about it. (Apollo is too complex to setup and
> > > have
> > > > > high
> > > > > > learning cure  in using/maintaining).
> > > > > >
> > > > > > I tried to build a config-center around Etcd for Java-Chassis,
> it's
> > > quite
> > > > > > simple. Most of the features from the CC(CSE) can be achieved.
> > > > > >
> > > > > > From my point of view, using Etcd is good for small projects,
> show
> > > cases
> > > > > > and in development, because we already have an Etcd instance
> > > > > > running(service-center needs it).
> > > > > >
> > > > > > The code is here(it's just a demonstration, needs improvement):
> > > > > > https://github.com/yangbor/servicecomb-java-chassis/tree/cc-etcd
> > > > > >
> > > > > > Any thoughts?
> > > > > >
> > > > > > --
> > > > > > Best Regards,
> > > > > > Yang.
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Best Regards,
> > > > > Yang.
> > > > >
> > >
> >
> >
> > --
> > Best Regards,
> > Yang.
> >



-- 
Best Regards,
Yang.

Reply via email to