2022-09-02 01:52:49 UTC - Ben Carver: So, I defined a custom _kind_ called `genericnamenode10`. I can see that a bunch of these pods are being created due to my reactive scaling configuration. That being said, there are _far_ more `prewarm-genericnamenode10` pods than actual `namenodeX` pods. Does this indicate that something is wrong? (Like, the `genericnamenode10` pre-warm pods are not actually being used?) https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1662083569694609?thread_ts=1662083569.694609&cid=C3TPCAQG1 ---- 2022-09-02 20:24:28 UTC - Brendan Doyle: do people have any recommendations for ec2 profiles for invokers. I'm kind of curious on people's opinions on tradeoffs and what type of ec2's should be leveraged for standard workloads since there's so many options these days? Is it better to horizontally scale invokers vs. vertically scale etc. I.e. is it better to go to 100gb ram hosts from 15 or 30gb ram or just scale out 5 more invokers. Or if anyone has done any research on this. I feel like I asked something similar a while back. Obviously if you vertically scale and don't scale cpu proportionally with it then you're going to be giving less cpu to each function container https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1662150268525019 ----
