Issue #789 has been updated by luke.

Status changed from Needs design decision to Rejected
Affected version set to 0.24.4

As DavidS says, this is currently better done using resource purging, and the 
long term correct solution is just to have the client track changes to the 
resource catalog, and then have some general policy about what those changes 
mean.

E.g., we could track when resources were in the old catalog but are not in the 
new catalog.  We could then have a policy that said, purge any resource that we 
remove from our catalog.

These are two relatively straightforward steps, and involve less work all 
around.
----------------------------------------
Feature #789: Configurable automatic object's purging
http://projects.reductivelabs.com/issues/show/789

Author: tim
Status: Rejected
Priority: Normal
Assigned to: luke
Category: unknown
Target version: unplanned
Complexity: Unknown
Patch: None
Affected version: 0.24.4
Keywords: 


While talking on IRC, Spike and I figured it could be worth while to make this 
a feature request.

The problem:

When you delete an object in your manifest, it's not handled anymore on the 
client. The best way at present to make it also delete the object on the client 
is to leave it in the manifest with some kind of "ensure => absent" 
declaration. Although this works in a lot of places, it can be very ugly and 
inconvenient, especially when you're building a module.

A solution:

Allow for a "control" parameter on object basis. Make it best practice to have 
this always default to something like "not-puppet". When the parameter is set 
to "puppet" *manually in the manifest*, allow puppet to run some pre-determined 
defined command to remove it. This would mean that puppet would need to check 
on each run if the list of available "puppet-controlled" objects has been 
changed in comparison with the last puppet run. If so, if any has been removed, 
run the pre-determined defined method with the name of the object to actually 
remove it.

Problems:

Lots. Even when not counting on sysadmin messing things up. Cache problems 
between runs could be one problem. I personally think that most problems could 
be solved by simply using sane defaults: If anything goes wrong, yell, but 
don't change anything.

I'd be happy if this feature was only available for puppet modules, but I can 
imagine further widespread usage of this. You also might want a global setting 
that simply disables this behaviour, even when control => puppet is explicitly 
set.


----------------------------------------
You have received this notification because you have either subscribed to it, 
or are involved in it.
To change your notification preferences, please click here: 
http://reductivelabs.com/redmine/my/account

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"Puppet Bugs" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/puppet-bugs?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to