Re: Charmers membership request!

[Cory Johns]

And my axe!  Er, I mean, +1!

On Wed, Jun 1, 2016 at 3:32 AM, Konstantinos Tsakalozos <
kos.tsakalo...@canonical.com> wrote:

> Thank you both for the vote of confidence. Lets go charming!
>
> On Tue, May 31, 2016 at 11:32 PM, Kevin Monroe  > wrote:
>
>> Since last December, I've been on 18 tours through the Review Queue with
>> Kostas.  We've seen the good, the bad, and a little bit of ugly together.
>> We're in such a harmony now that his 'ack' is really just busy work for a
>> ~charmer to come behind and mash the buttons.
>>
>> Couple his dedication to the rev queue with his Big Data charm
>> contributions, and this is an easy +1 for me.  Thanks for the application
>> Kostas; you'll be a great addition to ~charmers.
>> --
>> Kevin Monroe
>>
>> On Tue, May 31, 2016 at 12:21 PM, Charles Butler <
>> charles.but...@canonical.com> wrote:
>>
>>> A big +1 to this application. You keep me honest, and always make me
>>> contribute higher quality submissions. Even if you have to lend a hand
>>> where my python-fu falls over. I look forward to having you join the ranks
>>> as a Charmer Konstantinos!
>>>
>>> On Mon, May 30, 2016 at 11:33 AM Konstantinos Tsakalozos <
>>> kos.tsakalo...@canonical.com> wrote:
>>>
 Hi,

 This is to request membership to ~charmers.

 I have been working with the Bigdata team for some time now. You can
 find my contributions in the respective charms & bundles on https://
 github.com/juju-solutions. Also, I've been taking part in the regular (
 normally weekly) review sessions of the team so I am very familiar
 with the process.

 Thank you,
 Konstantinos Tsakalozos
 --
 Juju mailing list
 Juju@lists.ubuntu.com
 Modify settings or unsubscribe at:
 https://lists.ubuntu.com/mailman/listinfo/juju

>>> --
>>> Juju Charmer
>>> Canonical Group Ltd.
>>> Ubuntu - Linux for human beings | www.ubuntu.com
>>> Juju - The fastest way to model your service | www.jujucharms.com
>>>
>>> --
>>> Juju mailing list
>>> Juju@lists.ubuntu.com
>>> Modify settings or unsubscribe at:
>>> https://lists.ubuntu.com/mailman/listinfo/juju
>>>
>>>
>>
>
> --
> Juju mailing list
> Juju@lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/juju
>
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

Re: Apache Drill Charm

[Merlijn Sebrechts]

Yeah, we should see if we can do the same with YARN...

2016-06-01 14:42 GMT+02:00 Marco Ceppi :

> +1 to using % logic to make it scale across any sized instance. Awesome
> stuff!
>
> On Wed, Jun 1, 2016 at 6:28 AM Tom Barber  wrote:
>
>> Okay latest "stable" build has RAM config options.
>>
>> Drill ships with defaults of 8GB and 3GB but I didn't want it to die on
>> EC2 Large etc boxes that dont have that much. So I added a bit of logic,
>> you can (I hope) add XXG and it will use that fixed amount, or you can, as
>> it ships, tell it you want XX% MAX and XX% Heap and it will try and figure
>> that out and stand you up a drill box.
>>
>>
>> Tom
>>
>> --
>>
>> Director Meteorite.bi - Saiku Analytics Founder
>> Tel: +44(0)5603641316
>>
>> (Thanks to the Saiku community we reached our Kickstart
>> 
>> goal, but you can always help by sponsoring the project
>> )
>>
>> On 1 June 2016 at 00:50, Tom Barber  wrote:
>>
>>> Oh, also currently the RAM is clamped down real low
>>> in /opt/drill/conf/drill-env.sh I will set it back to some sane defaults
>>> tomorrow as soon as I put the RAM limits into the config options, just ran
>>> out of time this evening!
>>>
>>> --
>>>
>>> Director Meteorite.bi - Saiku Analytics Founder
>>> Tel: +44(0)5603641316
>>>
>>> (Thanks to the Saiku community we reached our Kickstart
>>> 
>>> goal, but you can always help by sponsoring the project
>>> )
>>>
>>> On 31 May 2016 at 23:50, Tom Barber  wrote:
>>>
 Here we are then, for Merlijn and anyone else interested in SQL
 interfaces to big data/NOSQL stuff.

 This is less than a days effort, so its patchy at best:

 https://jujucharms.com/u/spicule/drillbit

 For those of you who don't know Apache Drill, it will let you run SQL
 querys over, CSV/JSON data, MongoDB, HBase, Parquet files etc in a number
 of different locations. Basically its a great way for analysts who use
 "traditional" SQL tools to leverage data stored within NOSQL solutions.

 Getting something like this into the CS has been high on my list of
 priorities for Saiku Analytics as it suddenly offers up loads of new
 connection prospects(of course I can do this manually in the past, but this
 is what Juju is for, right?)

 You need to deploy a ZK node (or 3) and connect it to that and OpenJDK
 to run it. Currently its relations-lite, the only one in there is a MongoDB
 test relation that will connect Drill to your MongoDB cluster if you run
 one, but there will be actions and relations coming shortly for other
 stuff. Also its missing a fat load of config options, again, coming soon.
 You can set all of this stuff pretty simply though and there is a web
 console for queries/connections etc, on http://serverip:8047/

 Test it, let me know what you think or its its entirely broken. Adding
 this and a few other NOSQL SQL interfaces is key to people being able to
 sanely consume all this big data stuff that is great on Juju. Thats not a
 knock on the Zeppelin guys, but companies will continue to use SQL for a
 long time to come so we should service that requirement.

 Tom

 --

 Director Meteorite.bi - Saiku Analytics Founder
 Tel: +44(0)5603641316

 (Thanks to the Saiku community we reached our Kickstart
 
 goal, but you can always help by sponsoring the project
 )

>>>
>>>
>> --
>> Juju mailing list
>> Juju@lists.ubuntu.com
>> Modify settings or unsubscribe at:
>> https://lists.ubuntu.com/mailman/listinfo/juju
>>
>
> --
> Juju mailing list
> Juju@lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/juju
>
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

Re: Apache Drill Charm

[Marco Ceppi]

+1 to using % logic to make it scale across any sized instance. Awesome
stuff!

On Wed, Jun 1, 2016 at 6:28 AM Tom Barber  wrote:

> Okay latest "stable" build has RAM config options.
>
> Drill ships with defaults of 8GB and 3GB but I didn't want it to die on
> EC2 Large etc boxes that dont have that much. So I added a bit of logic,
> you can (I hope) add XXG and it will use that fixed amount, or you can, as
> it ships, tell it you want XX% MAX and XX% Heap and it will try and figure
> that out and stand you up a drill box.
>
>
> Tom
>
> --
>
> Director Meteorite.bi - Saiku Analytics Founder
> Tel: +44(0)5603641316
>
> (Thanks to the Saiku community we reached our Kickstart
> 
> goal, but you can always help by sponsoring the project
> )
>
> On 1 June 2016 at 00:50, Tom Barber  wrote:
>
>> Oh, also currently the RAM is clamped down real low
>> in /opt/drill/conf/drill-env.sh I will set it back to some sane defaults
>> tomorrow as soon as I put the RAM limits into the config options, just ran
>> out of time this evening!
>>
>> --
>>
>> Director Meteorite.bi - Saiku Analytics Founder
>> Tel: +44(0)5603641316
>>
>> (Thanks to the Saiku community we reached our Kickstart
>> 
>> goal, but you can always help by sponsoring the project
>> )
>>
>> On 31 May 2016 at 23:50, Tom Barber  wrote:
>>
>>> Here we are then, for Merlijn and anyone else interested in SQL
>>> interfaces to big data/NOSQL stuff.
>>>
>>> This is less than a days effort, so its patchy at best:
>>>
>>> https://jujucharms.com/u/spicule/drillbit
>>>
>>> For those of you who don't know Apache Drill, it will let you run SQL
>>> querys over, CSV/JSON data, MongoDB, HBase, Parquet files etc in a number
>>> of different locations. Basically its a great way for analysts who use
>>> "traditional" SQL tools to leverage data stored within NOSQL solutions.
>>>
>>> Getting something like this into the CS has been high on my list of
>>> priorities for Saiku Analytics as it suddenly offers up loads of new
>>> connection prospects(of course I can do this manually in the past, but this
>>> is what Juju is for, right?)
>>>
>>> You need to deploy a ZK node (or 3) and connect it to that and OpenJDK
>>> to run it. Currently its relations-lite, the only one in there is a MongoDB
>>> test relation that will connect Drill to your MongoDB cluster if you run
>>> one, but there will be actions and relations coming shortly for other
>>> stuff. Also its missing a fat load of config options, again, coming soon.
>>> You can set all of this stuff pretty simply though and there is a web
>>> console for queries/connections etc, on http://serverip:8047/
>>>
>>> Test it, let me know what you think or its its entirely broken. Adding
>>> this and a few other NOSQL SQL interfaces is key to people being able to
>>> sanely consume all this big data stuff that is great on Juju. Thats not a
>>> knock on the Zeppelin guys, but companies will continue to use SQL for a
>>> long time to come so we should service that requirement.
>>>
>>> Tom
>>>
>>> --
>>>
>>> Director Meteorite.bi - Saiku Analytics Founder
>>> Tel: +44(0)5603641316
>>>
>>> (Thanks to the Saiku community we reached our Kickstart
>>> 
>>> goal, but you can always help by sponsoring the project
>>> )
>>>
>>
>>
> --
> Juju mailing list
> Juju@lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/juju
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

Re: Apache Drill Charm

[Tom Barber]

Okay latest "stable" build has RAM config options.

Drill ships with defaults of 8GB and 3GB but I didn't want it to die on EC2
Large etc boxes that dont have that much. So I added a bit of logic, you
can (I hope) add XXG and it will use that fixed amount, or you can, as it
ships, tell it you want XX% MAX and XX% Heap and it will try and figure
that out and stand you up a drill box.


Tom

--

Director Meteorite.bi - Saiku Analytics Founder
Tel: +44(0)5603641316

(Thanks to the Saiku community we reached our Kickstart

goal, but you can always help by sponsoring the project
)

On 1 June 2016 at 00:50, Tom Barber  wrote:

> Oh, also currently the RAM is clamped down real low
> in /opt/drill/conf/drill-env.sh I will set it back to some sane defaults
> tomorrow as soon as I put the RAM limits into the config options, just ran
> out of time this evening!
>
> --
>
> Director Meteorite.bi - Saiku Analytics Founder
> Tel: +44(0)5603641316
>
> (Thanks to the Saiku community we reached our Kickstart
> 
> goal, but you can always help by sponsoring the project
> )
>
> On 31 May 2016 at 23:50, Tom Barber  wrote:
>
>> Here we are then, for Merlijn and anyone else interested in SQL
>> interfaces to big data/NOSQL stuff.
>>
>> This is less than a days effort, so its patchy at best:
>>
>> https://jujucharms.com/u/spicule/drillbit
>>
>> For those of you who don't know Apache Drill, it will let you run SQL
>> querys over, CSV/JSON data, MongoDB, HBase, Parquet files etc in a number
>> of different locations. Basically its a great way for analysts who use
>> "traditional" SQL tools to leverage data stored within NOSQL solutions.
>>
>> Getting something like this into the CS has been high on my list of
>> priorities for Saiku Analytics as it suddenly offers up loads of new
>> connection prospects(of course I can do this manually in the past, but this
>> is what Juju is for, right?)
>>
>> You need to deploy a ZK node (or 3) and connect it to that and OpenJDK to
>> run it. Currently its relations-lite, the only one in there is a MongoDB
>> test relation that will connect Drill to your MongoDB cluster if you run
>> one, but there will be actions and relations coming shortly for other
>> stuff. Also its missing a fat load of config options, again, coming soon.
>> You can set all of this stuff pretty simply though and there is a web
>> console for queries/connections etc, on http://serverip:8047/
>>
>> Test it, let me know what you think or its its entirely broken. Adding
>> this and a few other NOSQL SQL interfaces is key to people being able to
>> sanely consume all this big data stuff that is great on Juju. Thats not a
>> knock on the Zeppelin guys, but companies will continue to use SQL for a
>> long time to come so we should service that requirement.
>>
>> Tom
>>
>> --
>>
>> Director Meteorite.bi - Saiku Analytics Founder
>> Tel: +44(0)5603641316
>>
>> (Thanks to the Saiku community we reached our Kickstart
>> 
>> goal, but you can always help by sponsoring the project
>> )
>>
>
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

Re: User Mgmt - 2.0 and Beyond

[Mark Shuttleworth]

Yes, this is super interesting. Thanks!


On 01/06/16 10:58, James Beedy wrote:
> I've "conjured-up" what I think would be a great enhancement to the current 
> user mgmt capability in 2.0. 
>
> As things stand, one can add a user to a model, and set a permission category 
> of either read (read), or write (read/write). This functionality is awesome, 
> and a huge step for juju (applause)! 
>
> Admins of juju can now create, manage, and maintain the users and users 
> access policy associated with a model (applause, again, seriously). 
>
> As a logical next step, why don't we take the user all the way to the 
> instance?
>
> What I'm thinking of is an '--os' flag that could be specified on user 
> creation!
>
> This flag would signify that the user need be created on the instances in the 
> current model. Ssh keys key(s) for a user could be added, and *associated*, 
> and provisioned alongside the respective user, and user account on the 
> machine.
>
> This functionality would give juju deployed infrastructure a huge edge in the 
> ease of user management/maintainability for any organization, and massive 
> bragging rights in enterprise land due to the increased PCI compliance 
> revolving around finer granularity in user access accounts.
>
> I feel like the majority of the big pieces are already In place, the primary 
> road blocks I foresee (probably a lot more):
> 1. User sensitive ssh-keys
> 2. Machine-level user provisioning template /UserManagerModel 
> 3. Os-level user access/permission policy (what is generic/default yet tuned 
> and hardened?)
>
>
> That about wraps it up, hopefully I got my point across to some degree.
>
> Thoughts?
>
>
>


-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

Re: User Mgmt - 2.0 and Beyond

[Samuel Cozannet]

Good start. I would add

* external AAA connectivity (LDAP/AD) to find where the actual data is
stored.
* add user to units only (no access to Juju)
* share

++
Sam


--
Samuel Cozannet
Cloud, Big Data and IoT Strategy Team
Business Development - Cloud and ISV Ecosystem
Changing the Future of Cloud
Ubuntu   / Canonical UK LTD  / Juju

samuel.cozan...@canonical.com
mob: +33 616 702 389
skype: samnco
Twitter: @SaMnCo_23
[image: View Samuel Cozannet's profile on LinkedIn]


On Wed, Jun 1, 2016 at 10:58 AM, James Beedy  wrote:

> I've "conjured-up" what I think would be a great enhancement to the
> current user mgmt capability in 2.0.
>
> As things stand, one can add a user to a model, and set a permission
> category of either read (read), or write (read/write). This functionality
> is awesome, and a huge step for juju (applause)!
>
> Admins of juju can now create, manage, and maintain the users and users
> access policy associated with a model (applause, again, seriously).
>
> As a logical next step, why don't we take the user all the way to the
> instance?
>
> What I'm thinking of is an '--os' flag that could be specified on user
> creation!
>
> This flag would signify that the user need be created on the instances in
> the current model. Ssh keys key(s) for a user could be added, and
> *associated*, and provisioned alongside the respective user, and user
> account on the machine.
>
> This functionality would give juju deployed infrastructure a huge edge in
> the ease of user management/maintainability for any organization, and
> massive bragging rights in enterprise land due to the increased PCI
> compliance revolving around finer granularity in user access accounts.
>
> I feel like the majority of the big pieces are already In place, the
> primary road blocks I foresee (probably a lot more):
> 1. User sensitive ssh-keys
> 2. Machine-level user provisioning template /UserManagerModel
> 3. Os-level user access/permission policy (what is generic/default yet
> tuned and hardened?)
>
>
> That about wraps it up, hopefully I got my point across to some degree.
>
> Thoughts?
>
>
>
> --
> Juju mailing list
> Juju@lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/juju
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

User Mgmt - 2.0 and Beyond

[James Beedy]

I've "conjured-up" what I think would be a great enhancement to the current 
user mgmt capability in 2.0. 

As things stand, one can add a user to a model, and set a permission category 
of either read (read), or write (read/write). This functionality is awesome, 
and a huge step for juju (applause)! 

Admins of juju can now create, manage, and maintain the users and users access 
policy associated with a model (applause, again, seriously). 

As a logical next step, why don't we take the user all the way to the instance?

What I'm thinking of is an '--os' flag that could be specified on user creation!

This flag would signify that the user need be created on the instances in the 
current model. Ssh keys key(s) for a user could be added, and *associated*, and 
provisioned alongside the respective user, and user account on the machine.

This functionality would give juju deployed infrastructure a huge edge in the 
ease of user management/maintainability for any organization, and massive 
bragging rights in enterprise land due to the increased PCI compliance 
revolving around finer granularity in user access accounts.

I feel like the majority of the big pieces are already In place, the primary 
road blocks I foresee (probably a lot more):
1. User sensitive ssh-keys
2. Machine-level user provisioning template /UserManagerModel 
3. Os-level user access/permission policy (what is generic/default yet tuned 
and hardened?)


That about wraps it up, hopefully I got my point across to some degree.

Thoughts?



-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

User Mgmt - 2.0 and Beyond

[James Beedy]

I've "conjured-up" what I think would be a great enhancement to the current 
user mgmt capability in 2.0. 

As things stand, one can add a user to a model, and set a permission category 
of either read (read), or write (read/write). This functionality is awesome, 
and a huge step for juju (applause)! 

Admins of juju can now create, manage, and maintain the users and users access 
policy associated with a model (applause, again, seriously). 

As a logical next step, why don't we take the user all the way to the instance?

What I'm thinking of is an '--os' flag that could be specified on user creation!

This flag would signify that the user need be created on the instances in the 
current model. Ssh keys key(s) for a user could be added, and *associated*, and 
provisioned alongside the respective user, and user account on the machine.

This functionality would give juju deployed infrastructure a huge edge in the 
ease of user management/maintainability for any organization, and massive 
bragging rights in enterprise land due to the increased PCI compliance 
revolving around finer granularity in user access accounts.

I feel like the majority of the big pieces are already In place, the primary 
road blocks I foresee (probably a lot more):
1. User sensitive ssh-keys
2. Machine-level user provisioning template /UserManagerModel 
3. Os-level user access/permission policy (what is generic/default yet tuned 
and hardened?)


That about wraps it up, hopefully I got my point across to some degree.

Thoughts?



-- 
Juju-dev mailing list
Juju-dev@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju-dev

Re: Anything providing Zookeeper?

[Tom Barber]

Yeah I saw that on the Drill list, apologies for not mentioning that
earlier! :)

RAM config in the charm will be available shortly, although I've got a
meeting in 5 minutes so it'll probably be an hour or so.

--

Director Meteorite.bi - Saiku Analytics Founder
Tel: +44(0)5603641316

(Thanks to the Saiku community we reached our Kickstart

goal, but you can always help by sponsoring the project
)

On 1 June 2016 at 09:51, Merlijn Sebrechts 
wrote:

> Thanks for this Charm! I've sent it to my colleagues who are working with
> Drill. We got a lot better performance by storing the data in Parquet
> format. Drill wasn't able to distribute the queries because the data was
> stored as json in HDFS.
>
>
>
> 2016-05-31 17:29 GMT+02:00 Tom Barber :
>
>> (PPS. Going back to our chat about drill speed the other week, I'm not
>> saying Drill is actually the most performant or best in terms of what
>> people need, but in terms of a tool to connect to lots of stuff and offer
>> pretty useful data federation, its top of my list. I do have Phoenix, Kylin
>> etc sat on the backburner because the should offer more performance but for
>> more specific usecases.)
>>
>> --
>>
>> Director Meteorite.bi - Saiku Analytics Founder
>> Tel: +44(0)5603641316
>>
>> (Thanks to the Saiku community we reached our Kickstart
>> 
>> goal, but you can always help by sponsoring the project
>> )
>>
>> On 31 May 2016 at 16:25, Tom Barber  wrote:
>>
>>> (I am pretty stoked about this stuff though because it finally brings
>>> the connectivity Saiku, Pentaho stuff, Spago etc can all leverage to sit on
>>> top of the Juju Big Data stuff sanely)
>>>
>>> --
>>>
>>> Director Meteorite.bi - Saiku Analytics Founder
>>> Tel: +44(0)5603641316
>>>
>>> (Thanks to the Saiku community we reached our Kickstart
>>> 
>>> goal, but you can always help by sponsoring the project
>>> )
>>>
>>> On 31 May 2016 at 16:21, Tom Barber  wrote:
>>>
 Sorry, forgot I mentioned that.

 There is an unstable drill charm currently because I'm hacking it
 around and pushing updates to the store whilst I test adding providers as
 hooks. Bear with me until tomorrow and I'll dump something into the stable
 channel. It does work, its very basic, missing auth and stuff, but the guts
 will spin you up a scalable drill cluster. I'll post some stuff tomorrow
 morning with details.

 Tom

 --

 Director Meteorite.bi - Saiku Analytics Founder
 Tel: +44(0)5603641316

 (Thanks to the Saiku community we reached our Kickstart
 
 goal, but you can always help by sponsoring the project
 )

 On 31 May 2016 at 16:18, Merlijn Sebrechts  wrote:

> The Charms for a Drill cluster? Or did I misread that part?
>
> 2016-05-31 17:16 GMT+02:00 Tom Barber :
>
>> Which charms?
>>
>> --
>>
>> Director Meteorite.bi - Saiku Analytics Founder
>> Tel: +44(0)5603641316
>>
>> (Thanks to the Saiku community we reached our Kickstart
>> 
>> goal, but you can always help by sponsoring the project
>> )
>>
>> On 31 May 2016 at 16:16, Merlijn Sebrechts <
>> merlijn.sebrec...@gmail.com> wrote:
>>
>>> Cool! I'm interested in these Charms. Are they online somewhere?
>>>
>>> 2016-05-31 16:59 GMT+02:00 Tom Barber :
>>>
 Just to clear up this thread.

 Thanks to Kos we tracked the the condition, where Drill was setting
 its config before the whole quorum was available.

 For reference I ended up implementing something a bit like:

 https://github.com/juju-solutions/layer-apache-spark/blob/master/reactive/spark.py#L159

 Thanks guys, I now have a working basic Apache Drill cluster \o/

 Tom

 --

 Director Meteorite.bi - Saiku Analytics Founder
 Tel: +44(0)5603641316

 (Thanks to the Saiku community we reached our Kickstart
 

Re: Anything providing Zookeeper?

[Merlijn Sebrechts]

Thanks for this Charm! I've sent it to my colleagues who are working with
Drill. We got a lot better performance by storing the data in Parquet
format. Drill wasn't able to distribute the queries because the data was
stored as json in HDFS.



2016-05-31 17:29 GMT+02:00 Tom Barber :

> (PPS. Going back to our chat about drill speed the other week, I'm not
> saying Drill is actually the most performant or best in terms of what
> people need, but in terms of a tool to connect to lots of stuff and offer
> pretty useful data federation, its top of my list. I do have Phoenix, Kylin
> etc sat on the backburner because the should offer more performance but for
> more specific usecases.)
>
> --
>
> Director Meteorite.bi - Saiku Analytics Founder
> Tel: +44(0)5603641316
>
> (Thanks to the Saiku community we reached our Kickstart
> 
> goal, but you can always help by sponsoring the project
> )
>
> On 31 May 2016 at 16:25, Tom Barber  wrote:
>
>> (I am pretty stoked about this stuff though because it finally brings the
>> connectivity Saiku, Pentaho stuff, Spago etc can all leverage to sit on top
>> of the Juju Big Data stuff sanely)
>>
>> --
>>
>> Director Meteorite.bi - Saiku Analytics Founder
>> Tel: +44(0)5603641316
>>
>> (Thanks to the Saiku community we reached our Kickstart
>> 
>> goal, but you can always help by sponsoring the project
>> )
>>
>> On 31 May 2016 at 16:21, Tom Barber  wrote:
>>
>>> Sorry, forgot I mentioned that.
>>>
>>> There is an unstable drill charm currently because I'm hacking it around
>>> and pushing updates to the store whilst I test adding providers as hooks.
>>> Bear with me until tomorrow and I'll dump something into the stable
>>> channel. It does work, its very basic, missing auth and stuff, but the guts
>>> will spin you up a scalable drill cluster. I'll post some stuff tomorrow
>>> morning with details.
>>>
>>> Tom
>>>
>>> --
>>>
>>> Director Meteorite.bi - Saiku Analytics Founder
>>> Tel: +44(0)5603641316
>>>
>>> (Thanks to the Saiku community we reached our Kickstart
>>> 
>>> goal, but you can always help by sponsoring the project
>>> )
>>>
>>> On 31 May 2016 at 16:18, Merlijn Sebrechts 
>>> wrote:
>>>
 The Charms for a Drill cluster? Or did I misread that part?

 2016-05-31 17:16 GMT+02:00 Tom Barber :

> Which charms?
>
> --
>
> Director Meteorite.bi - Saiku Analytics Founder
> Tel: +44(0)5603641316
>
> (Thanks to the Saiku community we reached our Kickstart
> 
> goal, but you can always help by sponsoring the project
> )
>
> On 31 May 2016 at 16:16, Merlijn Sebrechts <
> merlijn.sebrec...@gmail.com> wrote:
>
>> Cool! I'm interested in these Charms. Are they online somewhere?
>>
>> 2016-05-31 16:59 GMT+02:00 Tom Barber :
>>
>>> Just to clear up this thread.
>>>
>>> Thanks to Kos we tracked the the condition, where Drill was setting
>>> its config before the whole quorum was available.
>>>
>>> For reference I ended up implementing something a bit like:
>>>
>>> https://github.com/juju-solutions/layer-apache-spark/blob/master/reactive/spark.py#L159
>>>
>>> Thanks guys, I now have a working basic Apache Drill cluster \o/
>>>
>>> Tom
>>>
>>> --
>>>
>>> Director Meteorite.bi - Saiku Analytics Founder
>>> Tel: +44(0)5603641316
>>>
>>> (Thanks to the Saiku community we reached our Kickstart
>>> 
>>> goal, but you can always help by sponsoring the project
>>> )
>>>
>>> On 31 May 2016 at 13:34, Konstantinos Tsakalozos <
>>> kos.tsakalo...@canonical.com> wrote:
>>>
 Hm... I do not see anything wrong with this part of the code.
 Looking at [0] you should be getting all three Zookeeper
 units/conversations.

 Would it be possible to share the charm you are building so that I
 can test it locally or ping us on IRC channel #juju at freenode for a 
 pair
 debugging session?


 [0]
 

Re: Subordinate charms

[Samuel Cozannet]

Hello,

What is then the recommendation for monitoring / logging / (...) agents
that run independently on the units and are loosely connected to the local
charms, but potentially still need to know about the apps running?

Wouldn´t colocation make them cumbersome? Still encoded as subordinate
charms they lack flexibility as they can´t change/be replaced.

For the Django example (and any framework sort of charms) I would have
thought that a layer is needed and the plugin is an option of the layer.

++
Sam





--
Samuel Cozannet
Cloud, Big Data and IoT Strategy Team
Business Development - Cloud and ISV Ecosystem
Changing the Future of Cloud
Ubuntu   / Canonical UK LTD  / Juju

samuel.cozan...@canonical.com
mob: +33 616 702 389
skype: samnco
Twitter: @SaMnCo_23
[image: View Samuel Cozannet's profile on LinkedIn]


On Tue, May 31, 2016 at 2:50 PM, Rick Harding 
wrote:

> Yes, this is the way that subordinates work. They can often do very
> intrusive things. Imagine a django subordinate that installs and sets up a
> plugin into the django application it's deployed to. We don't currently
> support the removal of them. They kind of become one with the principle
> application.
>
> On Thu, May 26, 2016 at 5:41 PM André Moreira  wrote:
>
>> Hi Tom,
>>
>> I found this:
>>
>> The current model of subordinates doesn't include support for removing
>> subordinate units from their principal service apart from removing the
>> principal service itself. This limitation stems from the current policy
>> around service shutdown and the invocation of stop hooks.
>> https://jujucharms.com/docs/1.25/authors-subordinate-services
>>
>> Is this policy still valid?
>>
>> André
>>
>> 2016-05-26 18:38 GMT-03:00 Tom Barber :
>>
>>> Hi Andre
>>>
>>> Can you give us a clue what you are installing/uninstalling because I
>>> believe Bilal is correct, I've not see it wipe out the parent charms either.
>>>
>>> Tom
>>>
>>> --
>>>
>>> Director Meteorite.bi - Saiku Analytics Founder
>>> Tel: +44(0)5603641316
>>>
>>> (Thanks to the Saiku community we reached our Kickstart
>>> 
>>> goal, but you can always help by sponsoring the project
>>> )
>>>
>>> On 26 May 2016 at 22:36, André Moreira  wrote:
>>>
 Using this, it also removes the unit of the charm it is subordinated to.

 2016-05-26 18:34 GMT-03:00 Bilal Baqar :

> Try the normal charm remove command:
> *juju remove-service  *
>
> Regards
>
> 2016-05-27 2:24 GMT+05:00 André Moreira :
>
>> How can I remove a subordinate charm without removing the principal?
>>
>> --
>>
>> Le doux charme de maint songe
>> Par leur bel art inventé
>> Sous les habits du mensonge
>> Nous offre la vérité.
>> -La Fontaine
>>
>> --
>> Juju mailing list
>> Juju@lists.ubuntu.com
>> Modify settings or unsubscribe at:
>> https://lists.ubuntu.com/mailman/listinfo/juju
>>
>>
>
>
> --
> Bilal Baqar
> MTS - PLUMgrid Inc.
>
>
>


 --

 Le doux charme de maint songe
 Par leur bel art inventé
 Sous les habits du mensonge
 Nous offre la vérité.
 -La Fontaine

 --
 Juju mailing list
 Juju@lists.ubuntu.com
 Modify settings or unsubscribe at:
 https://lists.ubuntu.com/mailman/listinfo/juju


>>>
>>
>>
>> --
>>
>> Le doux charme de maint songe
>> Par leur bel art inventé
>> Sous les habits du mensonge
>> Nous offre la vérité.
>> -La Fontaine
>> --
>> Juju mailing list
>> Juju@lists.ubuntu.com
>> Modify settings or unsubscribe at:
>> https://lists.ubuntu.com/mailman/listinfo/juju
>>
>
> --
> Juju mailing list
> Juju@lists.ubuntu.com
> Modify settings or unsubscribe at:
> https://lists.ubuntu.com/mailman/listinfo/juju
>
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju

Re: Charmers membership request!

[Konstantinos Tsakalozos]

Thank you both for the vote of confidence. Lets go charming!

On Tue, May 31, 2016 at 11:32 PM, Kevin Monroe 
wrote:

> Since last December, I've been on 18 tours through the Review Queue with
> Kostas.  We've seen the good, the bad, and a little bit of ugly together.
> We're in such a harmony now that his 'ack' is really just busy work for a
> ~charmer to come behind and mash the buttons.
>
> Couple his dedication to the rev queue with his Big Data charm
> contributions, and this is an easy +1 for me.  Thanks for the application
> Kostas; you'll be a great addition to ~charmers.
> --
> Kevin Monroe
>
> On Tue, May 31, 2016 at 12:21 PM, Charles Butler <
> charles.but...@canonical.com> wrote:
>
>> A big +1 to this application. You keep me honest, and always make me
>> contribute higher quality submissions. Even if you have to lend a hand
>> where my python-fu falls over. I look forward to having you join the ranks
>> as a Charmer Konstantinos!
>>
>> On Mon, May 30, 2016 at 11:33 AM Konstantinos Tsakalozos <
>> kos.tsakalo...@canonical.com> wrote:
>>
>>> Hi,
>>>
>>> This is to request membership to ~charmers.
>>>
>>> I have been working with the Bigdata team for some time now. You can
>>> find my contributions in the respective charms & bundles on https://
>>> github.com/juju-solutions. Also, I've been taking part in the regular (
>>> normally weekly) review sessions of the team so I am very familiar with
>>> the process.
>>>
>>> Thank you,
>>> Konstantinos Tsakalozos
>>> --
>>> Juju mailing list
>>> Juju@lists.ubuntu.com
>>> Modify settings or unsubscribe at:
>>> https://lists.ubuntu.com/mailman/listinfo/juju
>>>
>> --
>> Juju Charmer
>> Canonical Group Ltd.
>> Ubuntu - Linux for human beings | www.ubuntu.com
>> Juju - The fastest way to model your service | www.jujucharms.com
>>
>> --
>> Juju mailing list
>> Juju@lists.ubuntu.com
>> Modify settings or unsubscribe at:
>> https://lists.ubuntu.com/mailman/listinfo/juju
>>
>>
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju