[[chef-dev]] Re: [[chef-dev]] Re: [[chef-dev]] Re: [[chef-dev]] Re: [[chef-dev]] CHEF-2224

[Seth Falcon < >]

On Thu, Apr 14, 2011 at 3:52 PM, Adam Jacob 
<
 >
 wrote:
> Works for me.
>
> Adam
>
> On Thu, Apr 14, 2011 at 5:51 PM, Seth Chisamore 
> <
 >
>  wrote:
>> So as a compromise can we dump the attribute data out to /tmp like we do
>> with the stack trace now (/tmp/chef-stacktrace.out)?  Maybe we can just 
>> make
>> the output on a failure contain node attribute data + exception...ie
>> /tmp/chef-failure.dump.

Nice! :-)

Another thing we might consider is having logic that would detect
whether the the first boot json file has already been applied or not.
For example, perhaps after it gets applied and the node successfully
saved, it could be moved to first-boot-applied-TIMESTAMP.  The idea
being that you don't have to use -j.

And FWIW, I like the compromise as I think avoiding the intermediate
node save makes for a cleaner system -- especially the "what happens
if chef-client runs on my load-balancer when I have a half-saved node"
scenario.

+ seth

-- 
Seth Falcon | Senior Software Design Engineer | Opscode | @sfalcon