You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@drill.apache.org by Arjun Rao <sp...@gmail.com> on 2018/07/26 15:45:03 UTC
Apache Drill on Kubernetes
Hi,
I am new to this forum and am excited to use Drill. This might have been
discussed in the past but I wanted to know if Apache Drill can be run on
Kubernetes and if not, if it's on the roadmap for Drill?
Appreciate the help!
Best,
Arjun
Re: Apache Drill on Kubernetes
Posted by Abhishek Girish <ag...@apache.org>.
@Arjun, sure. I've not faced the need for any changes for drillbit
discovery. Internally at MapR, we are using Helm charts for deployment -
for both on-prem and on GKE. For on-prem, we use Weave Net for networking
(plus a few other configurations shared on the K8S docs). We've also made
sure to automate ZK discovery - which helps avoiding manual config. The
GitHub repo I shared only has some basic support for Apache Drill on K8S
(it should still work out of the box - with limited manual config). I'll
work on updating the repo soon.
And yes, as we make progress with the official support into Drill, your
help will be appreciated. Will be tracked by DRILL-6598
<https://issues.apache.org/jira/browse/DRILL-6598>
On Thu, Jul 26, 2018 at 6:26 PM Arjun Rao <sp...@gmail.com> wrote:
> Thanks for the responses. These are all very promising.
>
> @Saurabh - We are heavy users of Kubernetes and are considering adding
> Apache Drill to our ecosystem.
> @Abhishek - We actually got it to run in our cluster but had to modify the
> drillbit discovery code a little bit on a local fork of ours to enable a
> multinode drill installation. We have our own docker container/helm charts
> for this multinode deployment. Before we went further down that road, we
> wanted to see what the community was doing. I will take a look at the repo
> you linked and let you know if I have any questions. If there is anything I
> can contribute to help with the K8s work, please let me know what I can do.
>
> Best,
> Arjun
>
> On Thu, Jul 26, 2018 at 1:51 PM John Omernik <jo...@omernik.com> wrote:
>
> > Dril, in it's current form, has had all the features required to be run
> on
> > cluster managers with multi-tenacy. There are many caveats but mostly
> > realted to how things are configured, not limitations of Drill itself.
> The
> > work being done for Drill in containers and Drill on K8s not only
> provides
> > a roadmap of how to do it, but reviewing the yaml and docker files is a
> > great way to see how devs are approaching a proper multitenant capable
> > Drill cluster. This is exciting work!
> >
> > On Thu, Jul 26, 2018, 11:40 AM Abhishek Girish <ag...@apache.org>
> wrote:
> >
> > > Hey everyone,
> > >
> > > Like John and Saurabh mentioned, yes this is possible. We've been using
> > > Drill on Kubernetes for a while now. I have a draft of my work
> > > (Dockerfiles + YAML definitions) available in [1]. Drill should come up
> > > successfully in distributed mode (multiple Drillbits) under K8S. Please
> > > give it a try when you can and reach out to me if you need any
> > > clarifications.
> > >
> > > For 1.14.0, we added official support for Docker. For the next release,
> > i'm
> > > working on incorporating the K8S support into the Apache Drill repo.
> > >
> > > [1] https://github.com/Agirish/drill-containers
> > >
> > > On Thu, Jul 26, 2018 at 8:59 AM Saurabh Mahapatra <
> > > saurabhmahapatra94@gmail.com> wrote:
> > >
> > > > Hey Arjun,
> > > >
> > > > Is the need for kubernetes a top down requirement in your
> architecture?
> > > >
> > > > John is right when it comes to running Drill inside a container. But
> > > there
> > > > was some talk of addressing the other problem which is whether K8 can
> > be
> > > a
> > > > resource manager for multiple Drill clusters...an alternative to
> > > splitting
> > > > up interactive and batch loads. I know this has been done within a
> YARN
> > > > setting.
> > > >
> > > > Paul and Abhishek are more knowledgeable about this. Thoughts?
> > > >
> > > > Maybe a topic for this year’s conference?
> > > >
> > > > Thanks,
> > > > Saurabh
> > > >
> > > > Sent from my iPhone
> > > >
> > > >
> > > >
> > > > > On Jul 26, 2018, at 8:48 AM, John Omernik <jo...@omernik.com>
> wrote:
> > > > >
> > > > > Absolutely it can! I don't have a docker file for it handy but I
> know
> > > it
> > > > > works well!
> > > > >
> > > > >> On Thu, Jul 26, 2018, 10:45 AM Arjun Rao <sp...@gmail.com>
> > > > wrote:
> > > > >>
> > > > >> Hi,
> > > > >> I am new to this forum and am excited to use Drill. This might
> have
> > > been
> > > > >> discussed in the past but I wanted to know if Apache Drill can be
> > run
> > > on
> > > > >> Kubernetes and if not, if it's on the roadmap for Drill?
> > > > >>
> > > > >> Appreciate the help!
> > > > >>
> > > > >> Best,
> > > > >> Arjun
> > > > >>
> > > >
> > >
> >
>
Re: Apache Drill on Kubernetes
Posted by Arjun Rao <sp...@gmail.com>.
Thanks for the responses. These are all very promising.
@Saurabh - We are heavy users of Kubernetes and are considering adding
Apache Drill to our ecosystem.
@Abhishek - We actually got it to run in our cluster but had to modify the
drillbit discovery code a little bit on a local fork of ours to enable a
multinode drill installation. We have our own docker container/helm charts
for this multinode deployment. Before we went further down that road, we
wanted to see what the community was doing. I will take a look at the repo
you linked and let you know if I have any questions. If there is anything I
can contribute to help with the K8s work, please let me know what I can do.
Best,
Arjun
On Thu, Jul 26, 2018 at 1:51 PM John Omernik <jo...@omernik.com> wrote:
> Dril, in it's current form, has had all the features required to be run on
> cluster managers with multi-tenacy. There are many caveats but mostly
> realted to how things are configured, not limitations of Drill itself. The
> work being done for Drill in containers and Drill on K8s not only provides
> a roadmap of how to do it, but reviewing the yaml and docker files is a
> great way to see how devs are approaching a proper multitenant capable
> Drill cluster. This is exciting work!
>
> On Thu, Jul 26, 2018, 11:40 AM Abhishek Girish <ag...@apache.org> wrote:
>
> > Hey everyone,
> >
> > Like John and Saurabh mentioned, yes this is possible. We've been using
> > Drill on Kubernetes for a while now. I have a draft of my work
> > (Dockerfiles + YAML definitions) available in [1]. Drill should come up
> > successfully in distributed mode (multiple Drillbits) under K8S. Please
> > give it a try when you can and reach out to me if you need any
> > clarifications.
> >
> > For 1.14.0, we added official support for Docker. For the next release,
> i'm
> > working on incorporating the K8S support into the Apache Drill repo.
> >
> > [1] https://github.com/Agirish/drill-containers
> >
> > On Thu, Jul 26, 2018 at 8:59 AM Saurabh Mahapatra <
> > saurabhmahapatra94@gmail.com> wrote:
> >
> > > Hey Arjun,
> > >
> > > Is the need for kubernetes a top down requirement in your architecture?
> > >
> > > John is right when it comes to running Drill inside a container. But
> > there
> > > was some talk of addressing the other problem which is whether K8 can
> be
> > a
> > > resource manager for multiple Drill clusters...an alternative to
> > splitting
> > > up interactive and batch loads. I know this has been done within a YARN
> > > setting.
> > >
> > > Paul and Abhishek are more knowledgeable about this. Thoughts?
> > >
> > > Maybe a topic for this year’s conference?
> > >
> > > Thanks,
> > > Saurabh
> > >
> > > Sent from my iPhone
> > >
> > >
> > >
> > > > On Jul 26, 2018, at 8:48 AM, John Omernik <jo...@omernik.com> wrote:
> > > >
> > > > Absolutely it can! I don't have a docker file for it handy but I know
> > it
> > > > works well!
> > > >
> > > >> On Thu, Jul 26, 2018, 10:45 AM Arjun Rao <sp...@gmail.com>
> > > wrote:
> > > >>
> > > >> Hi,
> > > >> I am new to this forum and am excited to use Drill. This might have
> > been
> > > >> discussed in the past but I wanted to know if Apache Drill can be
> run
> > on
> > > >> Kubernetes and if not, if it's on the roadmap for Drill?
> > > >>
> > > >> Appreciate the help!
> > > >>
> > > >> Best,
> > > >> Arjun
> > > >>
> > >
> >
>
Re: Apache Drill on Kubernetes
Posted by John Omernik <jo...@omernik.com>.
Dril, in it's current form, has had all the features required to be run on
cluster managers with multi-tenacy. There are many caveats but mostly
realted to how things are configured, not limitations of Drill itself. The
work being done for Drill in containers and Drill on K8s not only provides
a roadmap of how to do it, but reviewing the yaml and docker files is a
great way to see how devs are approaching a proper multitenant capable
Drill cluster. This is exciting work!
On Thu, Jul 26, 2018, 11:40 AM Abhishek Girish <ag...@apache.org> wrote:
> Hey everyone,
>
> Like John and Saurabh mentioned, yes this is possible. We've been using
> Drill on Kubernetes for a while now. I have a draft of my work
> (Dockerfiles + YAML definitions) available in [1]. Drill should come up
> successfully in distributed mode (multiple Drillbits) under K8S. Please
> give it a try when you can and reach out to me if you need any
> clarifications.
>
> For 1.14.0, we added official support for Docker. For the next release, i'm
> working on incorporating the K8S support into the Apache Drill repo.
>
> [1] https://github.com/Agirish/drill-containers
>
> On Thu, Jul 26, 2018 at 8:59 AM Saurabh Mahapatra <
> saurabhmahapatra94@gmail.com> wrote:
>
> > Hey Arjun,
> >
> > Is the need for kubernetes a top down requirement in your architecture?
> >
> > John is right when it comes to running Drill inside a container. But
> there
> > was some talk of addressing the other problem which is whether K8 can be
> a
> > resource manager for multiple Drill clusters...an alternative to
> splitting
> > up interactive and batch loads. I know this has been done within a YARN
> > setting.
> >
> > Paul and Abhishek are more knowledgeable about this. Thoughts?
> >
> > Maybe a topic for this year’s conference?
> >
> > Thanks,
> > Saurabh
> >
> > Sent from my iPhone
> >
> >
> >
> > > On Jul 26, 2018, at 8:48 AM, John Omernik <jo...@omernik.com> wrote:
> > >
> > > Absolutely it can! I don't have a docker file for it handy but I know
> it
> > > works well!
> > >
> > >> On Thu, Jul 26, 2018, 10:45 AM Arjun Rao <sp...@gmail.com>
> > wrote:
> > >>
> > >> Hi,
> > >> I am new to this forum and am excited to use Drill. This might have
> been
> > >> discussed in the past but I wanted to know if Apache Drill can be run
> on
> > >> Kubernetes and if not, if it's on the roadmap for Drill?
> > >>
> > >> Appreciate the help!
> > >>
> > >> Best,
> > >> Arjun
> > >>
> >
>
Re: Apache Drill on Kubernetes
Posted by Abhishek Girish <ag...@apache.org>.
Hey everyone,
Like John and Saurabh mentioned, yes this is possible. We've been using
Drill on Kubernetes for a while now. I have a draft of my work
(Dockerfiles + YAML definitions) available in [1]. Drill should come up
successfully in distributed mode (multiple Drillbits) under K8S. Please
give it a try when you can and reach out to me if you need any
clarifications.
For 1.14.0, we added official support for Docker. For the next release, i'm
working on incorporating the K8S support into the Apache Drill repo.
[1] https://github.com/Agirish/drill-containers
On Thu, Jul 26, 2018 at 8:59 AM Saurabh Mahapatra <
saurabhmahapatra94@gmail.com> wrote:
> Hey Arjun,
>
> Is the need for kubernetes a top down requirement in your architecture?
>
> John is right when it comes to running Drill inside a container. But there
> was some talk of addressing the other problem which is whether K8 can be a
> resource manager for multiple Drill clusters...an alternative to splitting
> up interactive and batch loads. I know this has been done within a YARN
> setting.
>
> Paul and Abhishek are more knowledgeable about this. Thoughts?
>
> Maybe a topic for this year’s conference?
>
> Thanks,
> Saurabh
>
> Sent from my iPhone
>
>
>
> > On Jul 26, 2018, at 8:48 AM, John Omernik <jo...@omernik.com> wrote:
> >
> > Absolutely it can! I don't have a docker file for it handy but I know it
> > works well!
> >
> >> On Thu, Jul 26, 2018, 10:45 AM Arjun Rao <sp...@gmail.com>
> wrote:
> >>
> >> Hi,
> >> I am new to this forum and am excited to use Drill. This might have been
> >> discussed in the past but I wanted to know if Apache Drill can be run on
> >> Kubernetes and if not, if it's on the roadmap for Drill?
> >>
> >> Appreciate the help!
> >>
> >> Best,
> >> Arjun
> >>
>
Re: Apache Drill on Kubernetes
Posted by Saurabh Mahapatra <sa...@gmail.com>.
Hey Arjun,
Is the need for kubernetes a top down requirement in your architecture?
John is right when it comes to running Drill inside a container. But there was some talk of addressing the other problem which is whether K8 can be a resource manager for multiple Drill clusters...an alternative to splitting up interactive and batch loads. I know this has been done within a YARN setting.
Paul and Abhishek are more knowledgeable about this. Thoughts?
Maybe a topic for this year’s conference?
Thanks,
Saurabh
Sent from my iPhone
> On Jul 26, 2018, at 8:48 AM, John Omernik <jo...@omernik.com> wrote:
>
> Absolutely it can! I don't have a docker file for it handy but I know it
> works well!
>
>> On Thu, Jul 26, 2018, 10:45 AM Arjun Rao <sp...@gmail.com> wrote:
>>
>> Hi,
>> I am new to this forum and am excited to use Drill. This might have been
>> discussed in the past but I wanted to know if Apache Drill can be run on
>> Kubernetes and if not, if it's on the roadmap for Drill?
>>
>> Appreciate the help!
>>
>> Best,
>> Arjun
>>
Re: Apache Drill on Kubernetes
Posted by John Omernik <jo...@omernik.com>.
Absolutely it can! I don't have a docker file for it handy but I know it
works well!
On Thu, Jul 26, 2018, 10:45 AM Arjun Rao <sp...@gmail.com> wrote:
> Hi,
> I am new to this forum and am excited to use Drill. This might have been
> discussed in the past but I wanted to know if Apache Drill can be run on
> Kubernetes and if not, if it's on the roadmap for Drill?
>
> Appreciate the help!
>
> Best,
> Arjun
>