[chef] Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: 0.10.RC Windows:site-cookbooks doesn't work


Chronological Thread 
  • From: Edward Sargisson < >
  • To:
  • Subject: [chef] Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: 0.10.RC Windows:site-cookbooks doesn't work
  • Date: Thu, 28 Apr 2011 16:05:54 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:content-type; b=X+Dxk3bEJHUrEshc0p32jSiKiiWu2EErxYmi2X6ysur3LC7AlT55m3eVu0dXPRi82V eu60IYlAHkoavyiW2wp2ZSl5LGt41SSBm+Xiy+gsEa4EL5vm3dNsTHasJuyFg3rzx3Mi d/N291BDd3US7viJ+Ve9zwrF5lD2UlBT8RZLM=

Okay... so if I understand the intent then, to take my concrete example:
Business need: The wordpress recipe doesn't quite work for me because it publishes an apache wordpress.conf only for workpress and not to connect to Tomcat as my site needs.

So I should fork the wordpress recipe, make a change so that the generation of the wordpress.conf can be suppressed based on an attribute and put the resulting change into the repository you created at [0] below so that it can be picked up if considered useful.

Then my local recipe becomes:
* set the attribute to suppress wordpress.conf
* include_recipe wordpress
* generate the site specific <site>.conf

Am I understanding things?

Cheers,
Edward

On Thu, Apr 28, 2011 at 3:44 PM, Hedge Hog < "> > wrote:
On Fri, Apr 29, 2011 at 4:24 AM, Edward Sargisson < "> > wrote:
> Hi,
> I'll happily try those debug steps tonight. However, perhaps you could
> explain somewhere (here or point me at an existing page) about how you
> envision the use case being met in the future.
> i.e. I want to take an existing cookbook, customise it a bit, but still be
> able to pick up changes to the parent cookbook when they're made and I'm
> happy with them.
>

I think this is actually a very common need, and it is useful to see
if people can come up with innovate ways of addressing it or, at
least, maybe provide best practice recipes targeted at this sorts of
use cases.
Before you read the next sentence please resolve to read all that
follows.  So I set up [0].

Now the caveats.
 - The Cookbooks account is still in the do-not-fork phase but will
come out of that in the next week.
 - The master branch is used to track opscode/cookbooks and the 37s
branch should, eventually, be updated with 37s changes (subject to
demand, or more likely subject to pull requests)
 - The master branch is updated periodically but not yet on a
schedule, that scheduled-time should be obvious to all, and easily
converted to local time, so I'm thinking midnight GMT.
 - Github collaborators have not yet been solicited so, depending on
their number, any pull requests might sit for a while, but that is
fine - you have your fork in your site-cookbooks folder.
 - A best practice git-workflow has not been documented for people not
familiar with git, for those who are: please ensure your pull request
merges cleanly with the 'qa' branch, and document your reasons in the
commit message.  Remeber others are going to have to be convinced you
change is a) good/best Chef practice (point to a url describing this
is fine), b) good/best practice for the app being configured, likewise
pointing to some blog/doc URL should suffice, after all it is the
internet age, and you are that generation ;)
 - The cuken project will make using these repos both documentable
(see [1]), executable and painless (or less-painful), but that project
too is in raw-meat/alpha stage....
 - The cuken project should be usable when I have chef-server 0.10.x
bootstrap-able from a Vagrantfile, and there have been several chef
issues blocking that (see my earlier mail thread), currently there is
one issue blocking this, but who knows what lies beyond that.
 - The Bundler experiment failed, and should be considered dead or, at
best, in a cryogenic tank.  Not because the idea of resolving
dependencies via Bundler, or some other library, is bad, but because
Bundler's code base was so horrific to work with and bundler core
(indirect) have also explicitly indicated Bundler's many git repo
gem-source 'issues', are not pain points for them right now.

While this is still a wip I HTH?

[0]: http://github.com/cookbooks
[1]: http://relishapp.com  among others

> Concrete example for discussion purposes: I recently took the wordpress
> cookbook and had to modify it and the code around it in order to attach an
> EBS volume for the database, include the FeedBurner plugin and my theme,
> work with Apache and Tomcat so I can send tomcat traffic to tomcat and blog
> traffic to Wordpress, etc.
>
> Cheers,
> Edward
>
> On Wed, Apr 27, 2011 at 10:11 PM, Daniel DeLeo < "> > wrote:
>>
>> On Friday, April 22, 2011 at 7:01 PM, Tom Thomas wrote:
>>
>> On Fri, Apr 22, 2011 at 3:30 PM, Edward Sargisson < "> >
>> wrote:
>>
>> Hi Tom,
>> Yes I was targetting a specific Availability Zone. In order to bring my
>> production site back up I had to use Chef 0.9 and specify a region to bring
>> it up.
>>
>> I've just tried the same command with 0.10.0.RC.0 and it still fails in
>> exactly the same way.
>>
>> if the chef version change had an impact, then it wasn't due to AWS
>> issues.  I'll leave analysis on why Chef 0.9 v. 0.10.0.RC had an impact on
>> the issues' occurrence to someone more qualified than me.
>>
>>
>> For clarity, I'm running Knife on a Windows client to startup an Ubuntu
>> instance in the cloud. The link below indicates that WinRM is required on
>> the Windows servers - not the client.
>> So, would you still like me to try it?
>>
>> ah, well then, probably not worth trying based in that :)  sorry I wasn't
>> of any help.
>>
>>
>> Cheers,
>> Edward
>>
>> In general, I'd recommend you migrate away from using site-cookbooks to
>> stitch together modified cookbooks.
>> That said, to debug this issue, it'd help if you can add `pp
>> @cookbook_settings` between line 67 and 68 of
>> chef/cookbook/cookbook_version_loader.rb and look for anything that stands
>> out (you can find the file with `gem which
>> chef/cookbook/cookbook_version_loader`
>>
>> --
>> Dan DeLeo
>>
>
>



--
πόλλ' οἶδ ἀλώπηξ, ἀλλ' ἐχῖνος ἓν μέγα
[The fox knows many things, but the hedgehog knows one big thing.]
  Archilochus, Greek poet (c. 680 BC – c. 645 BC)
http://wiki.hedgehogshiatus.com




Archive powered by MHonArc 2.6.16.

§