Re: [Fab-user] question about Fabric as a library

[Jeff Forcier]

Yup, that's one thing I kinda regret about the original result-object
API - its occasional convenience is often outweighed by how confusing
it is. Version 2 won't be doing that, for sure :) Anyway, glad you got
it sorted!

On Fri, Sep 6, 2013 at 6:46 AM, Vincent Derelle
<address@hidden> wrote:
> You're right.  I did a print on the dict mapping object returned by execute() 
> which gave me {u'node1': 'the output of the task that ran for node1', 
> u'node2': 'the output of the task that ran for node2'}.
>
> Then I did a:
>
> foobar = results['node1']
> print foobar
>
> which displayed again the task output.
>
> I got confused by the "to_string" methods of the objects. Sorry about that, 
> I'm relatively new to Python.
>
> Thanks for the reply!
>
> Vincent
> -----Original Message-----
> From: address@hidden [mailto:address@hidden On Behalf Of Jeff Forcier
> Sent: 5 septembre 2013 19:30
> To: Vincent Derelle
> Cc: address@hidden
> Subject: Re: [Fab-user] question about Fabric as a library
>
> The values involved ought to be the same either way - have you verified this?
>
> * sudo() returns an _AttributeString object which is the one you mentioned as 
> having useful attributes.
> * execute() returns that dict mapping hostname to the return value of the 
> task function, for that hostname.
> * Your task is returning the result of the sudo() calls, meaning they should 
> end up in the result of execute().
>
> Therefore, you should be seeing e.g.:
>
>     {'node1': <_AttributeString object from the sudo()>, 'node2':
> <_AttributeString object from the sudo()>}
>
> Where those _AttributeStrings are the same values you would see if you ran 
> your 2nd example snippet.
>
> If you're seeing something different, clarification and specific output 
> examples would be great :)
>
> Thanks,
> Jeff
>
> On Thu, Sep 5, 2013 at 1:45 PM, Vincent Derelle <address@hidden> wrote:
>> Hello,
>>
>>
>>
>> I'm developing an internal tool in our organisation. In a nutshell,
>> it's a web app (using Python Django) which retrieves a set of nodes
>> from a Chef server and then run commands on those selected nodes. I'm
>> using Fabric
>>
>> to execute those commands remotely. The recommended way to use Fabric
>> as a library in the documentation is to use the execute
>> (http://docs.fabfile.org/en/1.7/api/core/tasks.html#fabric.tasks.execu
>> te)
>>
>> function.
>>
>>
>>
>> If I use the execute function like this:
>>
>> nodes = ['nodes1', 'node2']
>>
>>
>>
>> result = execute(_fabric_execute, command, hosts=nodes)
>>
>>
>>
>> @parallel
>>
>> def _fabric_execute(command):
>>
>>     result = sudo(command)
>>
>>     return result
>>
>>
>>
>> this will work and return to me a dictionary containing the output of
>> those commands.
>>
>>
>>
>> If I do it sequentially instead like this:
>>
>> nodes = ['nodes1', 'nodes2']
>>
>>
>>
>> results = []
>>
>>
>>
>> for n in nodes:
>>
>>     env.host_string = n
>>
>>     result = _fabric_execute(command);
>>
>>     results.append(result)
>>
>>
>>
>> (in the django template iterate through the array and then display the
>> different result object properties)
>>
>>
>>
>> The individual result objects returned give me much more valuable
>> information such as: result.succeeded if it succeeded , result.failed
>> if it failed, result.real_command for the real command executed as
>> well as the the output.
>>
>>
>>
>> My question to you is, is there any way I can get that information
>> when using the execute function? Because my web app allows people to
>> execute commands, by default, in parallel which returns only the
>> output but if they chose sequential execution but I'm able to display
>> to them if it succeeded or not and the real command executed by fabric
>> if they select sequential execution. Essentially I'd like to be able
>> to display the same amount of information (mainly if it failed or not)
>> in parallel execution using execute like sequential execution
>>
>>
>>
>> Cheers,
>>
>> I love fabric!
>>
>>
>>
>> Vincent
>>
>>
>>
>>
>> _______________________________________________
>> Fab-user mailing list
>> address@hidden
>> https://lists.nongnu.org/mailman/listinfo/fab-user
>>
>
>
>
> --
> Jeff Forcier
> Unix sysadmin; Python/Ruby engineer
> http://bitprophet.org



-- 
Jeff Forcier
Unix sysadmin; Python/Ruby engineer
http://bitprophet.org