Stephen Hahn wrote:
> * Stephen Hahn <sch at sun.com> [2008-09-10 16:16]:
>> * David Bustos <David.Bustos at sun.com> [2008-09-10 01:42]:
>>> Quoth Jordan Brown on Tue, Sep 09, 2008 at 12:40:01PM -0700:
>>>> David Bustos wrote:
>>>>> And this wouldn't be stable, since "svcs" output is unstable.
>>>> Is "svcs -H -o xxx,yyy,zzz" output unstable?  It really should be 
>>>> stable, since it's clearly intended for script consumption.
>>> Yes, it is unstable.  I don't think script suitability should be married
>>> to stability.  File an RFE, if you wish.
>>  
>>   svcprop(1) provides a stable interface to all of the data that svcs -o
>>   would retrieve; it's designed for scripting (although I think Nico has
>>   a planned revision that improves its ease of use in scripting
>>   scenarios).

Indeed.  Two questions:

1)  Is there a list of the stabilities of the various properties?  I 
assume that some are stable, but I assume that others are not.

2)  Where's the documentation for the full FMRI including property, e.g. 
"svc:/application/scn/dhcpd:default/:properties/start/exec"?

>     ... except I don't think we've a stable scripting interface for
>     retrieving all instances on the system.  That's an omission, so I'll
>     go add it to Jordan's RFE.

svcprop '*'

appears to list all of the properties for all of the instances in the 
system.  Variations involving -p seem able to trim that down a bit, and 
in particular "svcprop -p restarter/state '*'" seems very plausible.

The result isn't quite the same as svcs '*':

- svcs includes lrc: services.
- If you're not careful which property you pick, multi-line properties 
make svcprop output awkward.
- svcprop includes services that have properties outside their 
instances.  (Or something like that.  I'm not sure I understand the 
difference here.)

but it's pretty close.


Reply via email to