Alejandro Baez

Still a bit gutted by the #hashicorp move to #busl 1.1. I completely understand their reasoning, but dammit does that emit #FUD. 😔

I don't expect hashicorp to do anything ill advised. But it's not them I'm worried about. It's any goliath coming along to scoop them up. Too many times this happened with excellent software companies. Moving to a far worse direction due to the licensing. ⚰️

Looking at you #Oracle. 😡 Still upset at that one for like 1000th time.

Sep 4, 2023 at 5:09pmWeb
Show threadAlejandro BaezSep 4, 2023

I am re-visiting my #nomad clusters as a result. I know this doesn't impact me in any way. I don't sell what I've built with it. Nor do I compete with hashicorp either. 😅

But I also haven't been able to open source most of it due to proprietary constraints on where I built these things. It just falls into this limbo of code that can benefit, but can't be shared yet. A BUSL essentially with almost no time constraint. 🫠

Show threadCarlos  🇸🇪Sep 4, 2023
@zeab coincidentally, I recently started moving my setup to a simpler approach with #dockercompose with #Portainer
Sad that Mitchel Hashimoto hasn't said anything publicly (yet?) about the company's new direction: though it's not his company anymore, it is his legacy.
Anyway, I'm happy how my services are coming back online very quickly because of less moving parts.
#backtobasics
Show threadAlejandro BaezSep 4, 2023

@mongrelion I think hashimoto had very little to do with this move. Like you stated, it's his legacy.

#Portainer can help a lot with simplicity. Especially if you all in on containers.

Personally, I taken immense advantage #nomad is a distributed job orchestrator. First and foremost.

It's allowed to schedule jobs for containers, microvm, system containers, wasm, and systemd spawn. And that's "only" the stuff I messed with throughout the years. 🙃 All thanks to it's versatility. 😅

Show threadCarlos  🇸🇪Sep 5, 2023
@zeab no doubt it's a powerful tool that I would recommend to my customers but for simpler use cases where you only have one single beefy server a.k.a. pet a.k.a. snowflake, you don't need a distributed systems orchestrator.
It's been a great learning experience, but I am looking forward to having a few less moving parts in my cluster.
Show threadAlejandro BaezSep 5, 2023

@mongrelion totally. Never run a job orchestrator unless you need it. The complexity can get intense. 😅

Though at the same time, even in one machine, you may have a need. All depends on your workload and what you trying to do. Complexity may be less.

That said I wouldn't ever say run #k8s for one machine. 😅 Even though you can. Whereas, with #nomad, I tend to favor even for one machine. Again all depends on what you trying to do.