You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@whirr.apache.org by Andrei Savu <sa...@gmail.com> on 2011/03/31 12:22:34 UTC
Quick planning for 0.5.0-incubating
Hi guys,
How about doing some planning for 0.5.0?
Some questions I'm thinking about:
- Should we focus on adding more services or on adding more cloud providers?
- Are there any JIRAs that we need to address as soon as possible?
- What's your short list of changes for 0.5.0?
- Should we have some sort of strategy for promoting the project so
that we can attract different communities to add services to Whirr?
- Is this the right moment to remove the Python contrib or we should
wait for one more release?
In the near future I'm going to work on making Whirr more useful for
build artefacts testing (e.g. automatic local file uploads to blob
stores) and on Puppet integration.
If we are aiming to release 0.5.0 by the end of June and no one else
wants / has time to be a release manager I could do this one more
time. For the next 3 months I'm going to work mostly on Whirr.
Cheers,
-- Andrei Savu / andreisavu.ro
Re: Quick planning for 0.5.0-incubating
Posted by Lars George <la...@gmail.com>.
Hi Andrei,
My major concern will be full HBase support, for CDH, 0.90.x and
onwards, Hadoop Append and so on.
I like to hear from others though about your thoughts. I am happy to
discuss them but do not have a strong opinion at the moment. Adding
more services is a great idea, this will show the versatility and
raise interest. As for the cloud providers, what is left? Which one
are not not supported yet but you think should? Which have missing
features (if there are any missing ones)?
How is using the Python scripts? I would add a deprecation warning in
0.5.0 and then remove it in the 0.6.0. - though they are outdated it
does not really ask for a quick removal. Unless there is no one using
it and then it could be dropped right away, no?
I would love to volunteer as release manager but honestly am in the
trenches writing the HBase book. So I would then do the one after
that. I certainly +1 your suggestion and seeing the awesome job you
have done this release I would love to see you handle another one :)
Lars
On Thu, Mar 31, 2011 at 12:22 PM, Andrei Savu <sa...@gmail.com> wrote:
> Hi guys,
>
> How about doing some planning for 0.5.0?
>
> Some questions I'm thinking about:
> - Should we focus on adding more services or on adding more cloud providers?
> - Are there any JIRAs that we need to address as soon as possible?
> - What's your short list of changes for 0.5.0?
> - Should we have some sort of strategy for promoting the project so
> that we can attract different communities to add services to Whirr?
> - Is this the right moment to remove the Python contrib or we should
> wait for one more release?
>
> In the near future I'm going to work on making Whirr more useful for
> build artefacts testing (e.g. automatic local file uploads to blob
> stores) and on Puppet integration.
>
> If we are aiming to release 0.5.0 by the end of June and no one else
> wants / has time to be a release manager I could do this one more
> time. For the next 3 months I'm going to work mostly on Whirr.
>
> Cheers,
>
> -- Andrei Savu / andreisavu.ro
>
Re: Quick planning for 0.5.0-incubating
Posted by Ashish <pa...@gmail.com>.
On Thu, Mar 31, 2011 at 3:52 PM, Andrei Savu <sa...@gmail.com> wrote:
> Hi guys,
>
> How about doing some planning for 0.5.0?
+1, way to go ;)
>
> Some questions I'm thinking about:
> - Should we focus on adding more services or on adding more cloud providers?
IMHO, completing service can take priority. But again, the best way is
get the community feedback and decide.
> - Are there any JIRAs that we need to address as soon as possible?
> - What's your short list of changes for 0.5.0?
1. solidify documentation is one thing that I can think of. Agree that
whirr is simple to use, but who start using it
need helpful information to satisfy their needs like a person looking
for Cassandra related stuff would like it specific to it, rather
than having to move through complete doco. Probably, it would be as
simple as adding more samples in doco.
2. FAQ's need to be strong, and simple way to build them is
incrementally adding the resolution of issues discussed in ML's
3. Presentation page need some luv. BTW, your whirr presentation on
slideshare must be added to this page :)
> - Should we have some sort of strategy for promoting the project so
> that we can attract different communities to add services to Whirr?
blog/tweet, and probably plan a session in upcoming ApacheCon :)
> - Is this the right moment to remove the Python contrib or we should
> wait for one more release?
>
> In the near future I'm going to work on making Whirr more useful for
> build artefacts testing (e.g. automatic local file uploads to blob
> stores) and on Puppet integration.
>
> If we are aiming to release 0.5.0 by the end of June and no one else
> wants / has time to be a release manager I could do this one more
> time. For the next 3 months I'm going to work mostly on Whirr.
>
> Cheers,
>
> -- Andrei Savu / andreisavu.ro
thanks
ashish
Re: Quick planning for 0.5.0-incubating
Posted by Tom White <to...@gmail.com>.
On Sat, Apr 2, 2011 at 10:18 AM, Adrian Cole <fe...@gmail.com> wrote:
> Andrei,
>
> Thanks for organizing this! I have a few thoughts:
>
> - community -
> Encourage those who are active in whirr to help spread the word. For
> example, if there is an event or meetup on nsql-x, have a lightning talk
> there. If there is a book in progress, get a page or two on whirr. If
> there are books, papers, presentations, interviews available, link to them
> on our site. That's enough for now :)
>
> - features -
> It is hard to go back and sweep up after a change across a dozen
> services. I'd focus on core features like dynamic resizing,
> auto-calculation of hardware requirements, integration with tools like
> puppet, pallet, saving state of services (generic snapshotting to
> blobstore). This type of code requires attention over weeks, a harder
> problem than implementing services, and takes good advantage of your months
> to offer!
+1
I've got a few patches underway for improvements to the core that
should make it easier to write services. E.g.
WHIRR-245 - Clearly demarcate the user and service provider APIs
WHIRR-262 - Services should not have to do reverse DNS lookups
WHIRR-275 - Improve firewall API for services
These should be done soon while there are fewer services to update.
>
> - documentation -
> Tom has a nice slideshare on writing whirr services. This should turn
> into a wiki.
I've added the presentation here:
https://cwiki.apache.org/confluence/display/WHIRR/Presentations.
I've also added a Powered By page to the wiki (at Patrick's suggestion):
https://cwiki.apache.org/confluence/display/WHIRR/Powered+By
Please add your organization and/or application to the list if it
makes use of Whirr.
Cheers,
Tom
>
> More later!
> -Adrian
> On Mar 31, 2011 3:23 AM, "Andrei Savu" <sa...@gmail.com> wrote:
>
Re: Quick planning for 0.5.0-incubating
Posted by Lars George <la...@gmail.com>.
BTW, I have a small section on Whirr in the HBase book.
On Apr 2, 2011, at 19:18, Adrian Cole <fe...@gmail.com> wrote:
> Andrei,
>
> Thanks for organizing this! I have a few thoughts:
>
> - community -
> Encourage those who are active in whirr to help spread the word. For
> example, if there is an event or meetup on nsql-x, have a lightning talk
> there. If there is a book in progress, get a page or two on whirr. If
> there are books, papers, presentations, interviews available, link to them
> on our site. That's enough for now :)
>
> - features -
> It is hard to go back and sweep up after a change across a dozen
> services. I'd focus on core features like dynamic resizing,
> auto-calculation of hardware requirements, integration with tools like
> puppet, pallet, saving state of services (generic snapshotting to
> blobstore). This type of code requires attention over weeks, a harder
> problem than implementing services, and takes good advantage of your months
> to offer!
>
> - documentation -
> Tom has a nice slideshare on writing whirr services. This should turn
> into a wiki.
>
> More later!
> -Adrian
> On Mar 31, 2011 3:23 AM, "Andrei Savu" <sa...@gmail.com> wrote:
Re: Quick planning for 0.5.0-incubating
Posted by Adrian Cole <fe...@gmail.com>.
Andrei,
Thanks for organizing this! I have a few thoughts:
- community -
Encourage those who are active in whirr to help spread the word. For
example, if there is an event or meetup on nsql-x, have a lightning talk
there. If there is a book in progress, get a page or two on whirr. If
there are books, papers, presentations, interviews available, link to them
on our site. That's enough for now :)
- features -
It is hard to go back and sweep up after a change across a dozen
services. I'd focus on core features like dynamic resizing,
auto-calculation of hardware requirements, integration with tools like
puppet, pallet, saving state of services (generic snapshotting to
blobstore). This type of code requires attention over weeks, a harder
problem than implementing services, and takes good advantage of your months
to offer!
- documentation -
Tom has a nice slideshare on writing whirr services. This should turn
into a wiki.
More later!
-Adrian
On Mar 31, 2011 3:23 AM, "Andrei Savu" <sa...@gmail.com> wrote: