juju command progress ui

[Ian Booth]

On 28/02/13 02:23, Brandon Holtsclaw wrote:
> juju returns after the command successfully runs because by design its
> async , you would not want the console to block when your doing something
> like 'juju deploy -n 50 hadoop' ... waiting for 50 nodes to finish before i
> could do something else is not good.
> 

Agreed for the deploy case. I could have been clearer perhaps, but I was mainly
talking about bootstrap and the ambiguity caused by it seeming to all go ok but
in the end it is still broken.

> 'juju status' will tell you the status of each service, node, and machines.
> 

Sure, but as a user I don't want to have to poll by running juju status over and
over just to see when the bootstrap node is ready.

> ... as far as progress and debugging, I think what your looking for is
> 'juju debug-log'
> 
Sorry, I should have been clearer, I'm talking about Go juju where debug-log
doesn't exist yet.

> ... as far as ssh I think what your looking for is 'juju ssh service/node'
> e.g. 'juju ssh mysql/0'
> 

I don't think this works in Go juju yet - or at least it doesn't for me.

> IIRC this should all be covered in the Documentation pretty well, is there
> areas you can see to make it clearer/improvements ?
> 

This email was intended more from a Go juju perspective. As I understand it, we
are looking to improve on some things from Python juju and so I was throwing up
some ideas which (for me) seemed like usability issues which may or may not be
deemed worth addressing. eg since debug-log doesn't exist yet, is the pyjuju
implementation sufficient or do we need so supplement it with other feedback
mechanisms like the observer model proposed elsewhere?