You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by "张铎 (Duo Zhang)" <pa...@gmail.com> on 2021/04/09 04:24:12 UTC

Re: [DISCUSS] About tracing in HBase

Hi, all the sub tasks for HBASE-22120 has been resolved.

And in https://issues.apache.org/jira/browse/HBASE-23762, I added a simple
documentation on how to enable tracing in HBase.

PTAL at the big PR

https://github.com/apache/hbase/pull/2901

If no big concerns, I will start a vote later to merge the branch back to
master.

Thanks.



张铎(Duo Zhang) <pa...@gmail.com> 于2021年2月27日周六 下午5:56写道:

> https://github.com/open-telemetry/opentelemetry-java/releases/tag/v1.0.0
>
> 1.0.0 is out finally. Let me start working.
>
> 张铎(Duo Zhang) <pa...@gmail.com> 于2021年2月19日周五 下午1:50写道:
>
>>
>>
>> Stack <st...@duboce.net> 于2021年2月19日周五 上午9:17写道:
>>
>>> On Sun, Feb 14, 2021 at 7:26 AM 张铎(Duo Zhang) <pa...@gmail.com>
>>> wrote:
>>>
>>> > Some updates here, I've successfully set up a cluster with tracing
>>> enabled
>>> > in my company.
>>> >
>>> >
>>> Hurray!
>>>
>>>
>>> > But the problem is that, during opentelemetry 0.14 -> 0.15, the API has
>>> > been changed a lot, so we need to modify our code accordingly.
>>> >
>>> >
>>> Is the incompatible change in API allowed because they are not at 1.0.0
>>> do
>>> you think?
>>>
>>> (Wondering how 'stable' 1.0.0 API will be when it comes out...)
>>>
>> They are publishing new releases very quickly, now it is 0.17.0, and
>> there are plenty of breaking changes in every release...
>> Seems their goal is to make a clean API when 1.0.0 is out...
>>
>>>
>>>
>>>
>>> > The good news is that the opentelemetry community kept publishing new
>>> > releases quickly recently, so the final 1.0.0 RC is likely to come
>>> soon.
>>> >
>>> > Will keep an eye on the new releases, and once it is done, will open
>>> new
>>> > issues to modify our code base and finish HBASE-22120.
>>> >
>>> >
>>> Thanks,
>>> S
>>>
>>>
>>>
>>> > Thanks.
>>> >
>>> > Sean Busbey <bu...@apache.org> 于2021年1月23日周六 上午2:40写道:
>>> >
>>> > > One of the big gaps in our tracing stuff to date is a lack of
>>> > > documentation. Without some guidance to show folks how to use the
>>> > > tracing, it was always unclear how broken or working our tracing was
>>> > > in htrace. I'd like to see some kind of docs in the feature branch
>>> > > prior to merge.
>>> > >
>>> > > On Thu, Jan 21, 2021 at 11:43 PM Duo Zhang <zh...@apache.org>
>>> wrote:
>>> > > >
>>> > > > In HBASE-22120, I tried to use OpenTelemetry to replace HTrace in
>>> > HBase,
>>> > > > and it is basically done.
>>> > > >
>>> > > > https://github.com/apache/hbase/pull/2901
>>> > > >
>>> > > > I've filed HBASE-25526 for later improvement.
>>> > > >
>>> > > > The remaining works for HBASE-22120 are
>>> > > > 1. Enabling trace on a real cluster, to test whether the whole flow
>>> > > works.
>>> > > > 2. Test the performance impact.
>>> > > > 3. Documentation.
>>> > > >
>>> > > > Once the above works are done and there are no big concerns, I will
>>> > > start a
>>> > > > vote to merge HBASE-22120 back to master.
>>> > > >
>>> > > > Thanks.
>>> > >
>>> >
>>>
>>

Re: [DISCUSS] About tracing in HBase

Posted by "张铎 (Duo Zhang)" <pa...@gmail.com>.
At the cluster side, if the client does not pass any trace info to the
server, we will start a new root span to do tracing.

So there is no problem for an hbase2 client talking with hbase3 cluster.

Stack <st...@duboce.net> 于2021年4月19日周一 下午11:53写道:

> Nice work here Duo. I think we should get this in.
>
> Any new pictures of traces to entice?
>
> I don't think this will get in the way of an hbase2 client talking to an
> hbase3 cluster; what do you think?
>
> S
>
> On Thu, Apr 8, 2021 at 9:24 PM 张铎(Duo Zhang) <pa...@gmail.com>
> wrote:
>
> > Hi, all the sub tasks for HBASE-22120 has been resolved.
> >
> > And in https://issues.apache.org/jira/browse/HBASE-23762, I added a
> simple
> > documentation on how to enable tracing in HBase.
> >
> > PTAL at the big PR
> >
> > https://github.com/apache/hbase/pull/2901
> >
> > If no big concerns, I will start a vote later to merge the branch back to
> > master.
> >
> > Thanks.
> >
> >
> >
> > 张铎(Duo Zhang) <pa...@gmail.com> 于2021年2月27日周六 下午5:56写道:
> >
> > >
> https://github.com/open-telemetry/opentelemetry-java/releases/tag/v1.0.0
> > >
> > > 1.0.0 is out finally. Let me start working.
> > >
> > > 张铎(Duo Zhang) <pa...@gmail.com> 于2021年2月19日周五 下午1:50写道:
> > >
> > >>
> > >>
> > >> Stack <st...@duboce.net> 于2021年2月19日周五 上午9:17写道:
> > >>
> > >>> On Sun, Feb 14, 2021 at 7:26 AM 张铎(Duo Zhang) <palomino219@gmail.com
> >
> > >>> wrote:
> > >>>
> > >>> > Some updates here, I've successfully set up a cluster with tracing
> > >>> enabled
> > >>> > in my company.
> > >>> >
> > >>> >
> > >>> Hurray!
> > >>>
> > >>>
> > >>> > But the problem is that, during opentelemetry 0.14 -> 0.15, the API
> > has
> > >>> > been changed a lot, so we need to modify our code accordingly.
> > >>> >
> > >>> >
> > >>> Is the incompatible change in API allowed because they are not at
> 1.0.0
> > >>> do
> > >>> you think?
> > >>>
> > >>> (Wondering how 'stable' 1.0.0 API will be when it comes out...)
> > >>>
> > >> They are publishing new releases very quickly, now it is 0.17.0, and
> > >> there are plenty of breaking changes in every release...
> > >> Seems their goal is to make a clean API when 1.0.0 is out...
> > >>
> > >>>
> > >>>
> > >>>
> > >>> > The good news is that the opentelemetry community kept publishing
> new
> > >>> > releases quickly recently, so the final 1.0.0 RC is likely to come
> > >>> soon.
> > >>> >
> > >>> > Will keep an eye on the new releases, and once it is done, will
> open
> > >>> new
> > >>> > issues to modify our code base and finish HBASE-22120.
> > >>> >
> > >>> >
> > >>> Thanks,
> > >>> S
> > >>>
> > >>>
> > >>>
> > >>> > Thanks.
> > >>> >
> > >>> > Sean Busbey <bu...@apache.org> 于2021年1月23日周六 上午2:40写道:
> > >>> >
> > >>> > > One of the big gaps in our tracing stuff to date is a lack of
> > >>> > > documentation. Without some guidance to show folks how to use the
> > >>> > > tracing, it was always unclear how broken or working our tracing
> > was
> > >>> > > in htrace. I'd like to see some kind of docs in the feature
> branch
> > >>> > > prior to merge.
> > >>> > >
> > >>> > > On Thu, Jan 21, 2021 at 11:43 PM Duo Zhang <zh...@apache.org>
> > >>> wrote:
> > >>> > > >
> > >>> > > > In HBASE-22120, I tried to use OpenTelemetry to replace HTrace
> in
> > >>> > HBase,
> > >>> > > > and it is basically done.
> > >>> > > >
> > >>> > > > https://github.com/apache/hbase/pull/2901
> > >>> > > >
> > >>> > > > I've filed HBASE-25526 for later improvement.
> > >>> > > >
> > >>> > > > The remaining works for HBASE-22120 are
> > >>> > > > 1. Enabling trace on a real cluster, to test whether the whole
> > flow
> > >>> > > works.
> > >>> > > > 2. Test the performance impact.
> > >>> > > > 3. Documentation.
> > >>> > > >
> > >>> > > > Once the above works are done and there are no big concerns, I
> > will
> > >>> > > start a
> > >>> > > > vote to merge HBASE-22120 back to master.
> > >>> > > >
> > >>> > > > Thanks.
> > >>> > >
> > >>> >
> > >>>
> > >>
> >
>

Re: [DISCUSS] About tracing in HBase

Posted by Stack <st...@duboce.net>.
Nice work here Duo. I think we should get this in.

Any new pictures of traces to entice?

I don't think this will get in the way of an hbase2 client talking to an
hbase3 cluster; what do you think?

S

On Thu, Apr 8, 2021 at 9:24 PM 张铎(Duo Zhang) <pa...@gmail.com> wrote:

> Hi, all the sub tasks for HBASE-22120 has been resolved.
>
> And in https://issues.apache.org/jira/browse/HBASE-23762, I added a simple
> documentation on how to enable tracing in HBase.
>
> PTAL at the big PR
>
> https://github.com/apache/hbase/pull/2901
>
> If no big concerns, I will start a vote later to merge the branch back to
> master.
>
> Thanks.
>
>
>
> 张铎(Duo Zhang) <pa...@gmail.com> 于2021年2月27日周六 下午5:56写道:
>
> > https://github.com/open-telemetry/opentelemetry-java/releases/tag/v1.0.0
> >
> > 1.0.0 is out finally. Let me start working.
> >
> > 张铎(Duo Zhang) <pa...@gmail.com> 于2021年2月19日周五 下午1:50写道:
> >
> >>
> >>
> >> Stack <st...@duboce.net> 于2021年2月19日周五 上午9:17写道:
> >>
> >>> On Sun, Feb 14, 2021 at 7:26 AM 张铎(Duo Zhang) <pa...@gmail.com>
> >>> wrote:
> >>>
> >>> > Some updates here, I've successfully set up a cluster with tracing
> >>> enabled
> >>> > in my company.
> >>> >
> >>> >
> >>> Hurray!
> >>>
> >>>
> >>> > But the problem is that, during opentelemetry 0.14 -> 0.15, the API
> has
> >>> > been changed a lot, so we need to modify our code accordingly.
> >>> >
> >>> >
> >>> Is the incompatible change in API allowed because they are not at 1.0.0
> >>> do
> >>> you think?
> >>>
> >>> (Wondering how 'stable' 1.0.0 API will be when it comes out...)
> >>>
> >> They are publishing new releases very quickly, now it is 0.17.0, and
> >> there are plenty of breaking changes in every release...
> >> Seems their goal is to make a clean API when 1.0.0 is out...
> >>
> >>>
> >>>
> >>>
> >>> > The good news is that the opentelemetry community kept publishing new
> >>> > releases quickly recently, so the final 1.0.0 RC is likely to come
> >>> soon.
> >>> >
> >>> > Will keep an eye on the new releases, and once it is done, will open
> >>> new
> >>> > issues to modify our code base and finish HBASE-22120.
> >>> >
> >>> >
> >>> Thanks,
> >>> S
> >>>
> >>>
> >>>
> >>> > Thanks.
> >>> >
> >>> > Sean Busbey <bu...@apache.org> 于2021年1月23日周六 上午2:40写道:
> >>> >
> >>> > > One of the big gaps in our tracing stuff to date is a lack of
> >>> > > documentation. Without some guidance to show folks how to use the
> >>> > > tracing, it was always unclear how broken or working our tracing
> was
> >>> > > in htrace. I'd like to see some kind of docs in the feature branch
> >>> > > prior to merge.
> >>> > >
> >>> > > On Thu, Jan 21, 2021 at 11:43 PM Duo Zhang <zh...@apache.org>
> >>> wrote:
> >>> > > >
> >>> > > > In HBASE-22120, I tried to use OpenTelemetry to replace HTrace in
> >>> > HBase,
> >>> > > > and it is basically done.
> >>> > > >
> >>> > > > https://github.com/apache/hbase/pull/2901
> >>> > > >
> >>> > > > I've filed HBASE-25526 for later improvement.
> >>> > > >
> >>> > > > The remaining works for HBASE-22120 are
> >>> > > > 1. Enabling trace on a real cluster, to test whether the whole
> flow
> >>> > > works.
> >>> > > > 2. Test the performance impact.
> >>> > > > 3. Documentation.
> >>> > > >
> >>> > > > Once the above works are done and there are no big concerns, I
> will
> >>> > > start a
> >>> > > > vote to merge HBASE-22120 back to master.
> >>> > > >
> >>> > > > Thanks.
> >>> > >
> >>> >
> >>>
> >>
>