[12:19:02] <_joe_>	 cdanis: a specific file, or any file with a specific extension?
[13:26:04] <cdanis>	 _joe_: I'm not sure that all .schema files in the puppet repo are jsonschema, but I'm not sure that isn't true, either
[14:21:25] <volans>	 cdanis: "a way" or "an easy way"?
[14:26:04] <_joe_>	 cdanis: we could add it to the spec for the module where the schemas are
[14:26:33] <cdanis>	 volans: if it isn't easy that is also a problem ;)
[14:29:14] <_joe_>	 it is easy
[14:29:23] <volans>	 you can have a look at the one for python lint
[14:29:24] <_joe_>	 it's just... ruby
[14:31:12] <_joe_>	 cdanis:  yeah either we change the extension to jsonschema or we just validate the files in the conftool dir
[14:45:29] <cdanis>	 _joe_: is that best done as a puppet spec test, or as a new thing defined in the rakefile for that module?
[14:45:41] <cdanis>	 it does't really depend on the compiled catalog
[14:53:58] <_joe_>	 cdanis: yeah
[18:59:12] <ottomata>	 hm
[19:02:25] <bblack>	 hm?
[19:02:37] <ottomata>	 ha, was going to ask an ATS question but then started reading more puppet
[19:02:52] <ottomata>	 we want to reuse stream.wm.org for multiple backend services
[19:02:59] <ottomata>	 with path based routing
[19:03:50] <bblack>	 I suppose that's an option :)
[19:04:01] <bblack>	 is there a good reason to? are they related?
[19:04:21] <ottomata>	 kind of, ya.  we could use a new domain, but good luck finding a good name :p
[19:04:30] <ottomata>	 https://phabricator.wikimedia.org/T236386#5667746
[19:04:39] <bblack>	 it's hard when all the generic words like "stream" are taken :)
[19:05:36] <ottomata>	 they are kinda related though, we are setting up stream intake / produce endpoints, and eventstreams is consume / export endpoint
[19:06:17] <bblack>	 I think my primary concern would be the protocol stuff
[19:06:22] <ottomata>	 oh?
[19:06:34] <ottomata>	 https you mean?
[19:06:35] <bblack>	 the current stream.wm.o is "special" in that it's a rare case that isn't plain http reqs
[19:06:45] <bblack>	 it uses websockets or something similar
[19:06:48] <ottomata>	 no no
[19:06:49] <ottomata>	 it is hhttp
[19:06:50] <ottomata>	 http
[19:06:57] <ottomata>	 taht was RCStream that used websockets
[19:06:59] <ottomata>	 which is decommed
[19:07:08] <ottomata>	 its just an unending http req
[19:07:28] <bblack>	 well either way, it is special
[19:07:46] <bblack>	 we don't normally have unending requests (either we're special-casing this as a piped connection, or you're suffering some odd timeouts, etc)
[19:08:10] <ottomata>	 yeah it is pipe via varnish rn
[19:08:13] <bblack>	 it might be nice to keep such special handling distinct (in terms of service routing) from any more-normal API endpoint that doesn't have endless connections
[19:08:17] <ottomata>	 https://www.irccloud.com/pastebin/DDP3Lns3/
[19:08:50] <ottomata>	 its kind of distinct already, in that only the /v2/stream url is pipe
[19:08:53] <bblack>	 although what "distinct" means I guess is questionable here
[19:08:54] <ottomata>	 right?
[19:09:19] <mutante>	 there is .stream gTLD /me hides
[19:10:04] <bblack>	 :P
[19:10:29] <ottomata>	 still reading, hm.  it looks like the ATS configs will support path based routing/rewriting....the varnish ones look like they pick a directory based on domain only?
[19:10:40] <ottomata>	 oh no
[19:10:49] <ottomata>	 sorry, director: eventstreams
[19:10:50] <ottomata>	 in the subpaths
[19:10:51] <ottomata>	 ok.
[19:46:54] <cdanis>	 ottomata: this is for the client-side error reporting launch, right?  why not just call it 'errorlogging.wm.o' ?
[19:47:06] <ottomata>	 cdanis:  there will be other endpoints too
[19:47:09] <ottomata>	 eventgate-analtyics-external
[19:47:21] <cdanis>	 ah okay