Rudder

I would like that too. In our case we're dealing with user desktops. And some users will turn their desktops off regularly, at least when they go on holidays. :)

+1

Related to http://www.rudder-project.org/redmine/issues/4234 as it is a better solution to the same problem.

olivier gave some more explanation on IRC which I don;'t see in this ticket:

< coredumb> would be nice to have a dedicated group: "No compliance check" 
< coredumb> which shows only last compliance known

We are thinking to an evolution of Rudder that would make the "compliance" mode a by-node choice, and that the default mode will become a report-only one.

More preciselly, in the default mode, we will only report last know state for node, and don't check for missing reports, plus have an health-check of the node (something along the "say you are alive every X minutes/hours/days). Reports would be send only on a configuration change or in case or repaired / error.
That will allow to simply change the frequency of the rudder agent on nodes, to enormously reduce the quantity of messages send in the network, and to no mark nodes regularly deconected from the network as in error (as in fact, we really want to know what was the last state).

The compliance mode would be what is done today (await for exepected reports, and in case they don't come, mark the node in "no answer").

So, it is not exactly what is depicted here, but the goals seems to be aligns, so it seemed to make sense to talk about it.

Cuuld be nice if the "(something along the "say you are alive every X minutes/hours/days)" is not mandatory.

Olivier Mauras wrote:

Cuuld be nice if the "(something along the "say you are alive every X minutes/hours/days)" is not mandatory.

You're right and it was the idea, thanks to have clarify it

Nice then :)
Can we hope to have that in 2.10 ?

Also it would be nice to have an option to remove/disable the cron script.
It would make sense to not have cf-agent running all the time for some particular setups.

Any updates on this one?