I recall them switching licenses and feeling dismayed but even then I hadn't used redis for many years. I'm still a huge antirez fanboy. I loved reading his C code.

Kind of tangent, but before he left, I didn't like the approach they took to clustering - they originally went with a "smarter/fat client" and bucket approach (at least afaikk might've changed more recently) and I'd wished they choose shards that proxied to each other and consistent hashing. There are decent reasons to do what they did, but it's also unsurprising that backend engineers tend to find justifications for fat clients easily ... "the hard to get properties are your responsibility not mine ... if you'd just do your job better, we'd have the utopia I'm forecasting, gosh."