|
I'd like to have the ability to write something akin to a report processor - e.g. real time processing of Puppet data, without having to use Ruby.
Currently, I've got no choice but to poll PuppetDB over and over if I want to do that. Native report processors are real time, which is nice, but you loose all the goodness PDB gives you (automatic queuing, retrying failed reports, and all that stuff I'd rather not reimplement again for every report processor I write in Ruby).
Of course we can use the store report processor, but now the reports are stored on disk on the particular master they happen to run against, which creates its own set of problems. (Do I rsync them? Put them on a network share? Stream them all to a db myself...oh wait that's PuppetDB).
I do think we've got a sort of chicken-and-the-egg thing here, which is that if we had it, customers would use it, but since we don't, it doesn't even occur to them.
Basically it'd be great to be able to tell a customer: Yes you can get real time data about what Puppet is doing, and you can get it with Perl if you want to.
|
