[07:34:33] <jayme>	 inflatador: have you tried reaching out to the other flink people that worked on this (like g.modena - sorry for putting you on the spot :))?
[08:49:04] <elukey>	 inflatador: I added some thoughts, didn't find anything weird so far. One question though - you have "dse-k8s-services" under helmfile.d, and inside the rdf-streaming-updater dir values-ds-k8s-eqiad.yaml, is there a reason not to use values.yaml directly? (to avoid weird settings merged with each other)
[13:36:46] <inflatador>	 jayme: I did talk to g.modena ...his flink app is using the k8s native HA, we're using ZK instead (for reasons...I can go into details if interested)
[13:39:07] <inflatador>	 elukey thanks for digging thru the logs, I missed that part. I'm a little worried that if you run Flink in K8s, you have to use K8s-native HA, which we didn't want to do. However, the Flink docs don't say this explicitly
[13:40:22] <inflatador>	 that could also be a problem with values inheritance like you mentioned...will try to use values.yaml as you suggested
[13:46:02] <elukey>	 inflatador: np! I think that the value shouldn't be picked up, but from a quick check I cannot see why zookeeper shouldn't be used
[13:48:00] <inflatador>	 Agreed, it's a pretty weak theory. Didn't have time to work on my reproducer yesterday, hopefully will get to that today
[13:48:36] <inflatador>	 Once I have something, I'll probably hit the Flink mailing list, they're pretty responsive
[14:19:02] <jayme>	 inflatador: yeah I know. But AIUI he also did some research in the zookeeper area
[14:19:19] <jayme>	 until we figured out that our main zookeeper is not usable :)
[14:22:21] <elukey>	 it seems more a settings issue, maybe we are missing something trivial
[14:22:39] <elukey>	 the values.yaml status is surely something that need a fix/clarification
[14:55:11] <inflatador>	 jayme ACK, I have been working closely w/him