Re: [pcp] RFC - pmie "ruleset" extension

[Ken McDonell <kenj@xxxxxxxxxxxxxxxx>]

On 24/06/14 11:33, Mark Goodwin wrote:
> ...
> Not just messy, but rather inefficient and probably error prone too
> because despite the repeated negation, the postfix.queues.* metrics
> could change between the multiple rule evaluations (so more than one
> rule could trigger). Actually - it depends on whether pmie does one
> fetch per rule or one fetch per update cycle(?)
I think this is OK in the general case ... Frank's collapsing of the 
fetch groups makes it much more likely that multiple rules with the same 
sampling interval and metrics in common will see consistent values 
across all of the rules.
The ruleset builds a single expression tree to span all the rules, so 
this further enforces consistency for any metrics appearing in multiple 
rules and actions.