I finally made one upgrade as I slowly rebuild my homelab... networking in the 1L Dell MiniPCs. As it turns out, there are inexpensive Intel i226v 2.5Gbps network cards that use the m.2 A+E key found in nearly every miniPC I've seen so far, but the Dell 70x0 sytems in particular have the perfect spot to mount them. I now have an internal, dedicated 2.5Gbps NIC for the storage backhaul and can leave the 1Gbps for management and/or services. Up until now, I've been using Pluggable USB3.0 2.5Gbps NICs for the storage backhaul but the cable management has been a bit of a nightmare. https://a.co/d/2JW4ipl

@Brandon Lee So far, no issues with drivers. The card I linked is based on an Intel i226v chipset and was picked up right away. The Pluggable USB NICs use Realtek chipsets and while I didn't have any issues with those specifically in Proxmox either, I've had terrible luck with some of the onboard 1Gbps Realtek NICs in older versions of Proxmox. Stability has improved all around since Procmox 9.

I have a DC UPS that powers the modem, a PiHole appliance on a dedicated Pi Zero 2, a small switch and a VOIP box. This, in combination with a dedicated UPS for the router (UDM Pro), can keep a minimally functional network and phone system running for ~2 hours. - NAS and Prod Cluster on one 1500VA UPS - Backup NAS and Dev Cluster on another 1500VA UPS - Office Desk / Gaming PC / Monitors / Switches on another 1500VA UPS - Router and "critical services" Proxmox Host on another 1500VA UPS In the event of a power outage, the Backup NAS initiates shutdowns on everything in the home lab after 15 seconds. First the Dev cluster (if online... it usually isn't), then the Prod cluster, then the Main NAS, then the Backup NAS itself. These UPS units can keep the systems up for about 20 minutes on fresh batteries, but I don't take chances... if the outage is longer than 10 seconds, it's likely going to be too long for the runtime anyway so shutdown process starts right away. The gaming PC can shut itself down if it's running. (usually off anyway) The Router and critical services host currently stay running until the UPS dies. If I'm home, I plug the UPS into a portable battery box (Anker Solix C1000) to sustain those hosts for up to an additional 5 hours. If I'm not home, physics takes over. The hardest part right now is that all four UPS units have become synchronized on battery replacements and tend to all need batteries every 3-4 years. I've been slowly replacing the UPS itself where possible if they're using "Simulated Sine Wave" outputs as those UPS units tend to break Active PFC supplies if running for too long. However, these UPS's rarely deal with outages near me. I am far more likely to experience brown-outs and surges that cause the AVR features to kick in and that is where I extract value. Nothing in the home lab needs to be up 24/7, that's just the fun part. Equipment protection from power instability is the key for me.

@Matthias Neumeister This is the approach I took as well. YouTube is blocked at the DNS level with the exception of a handful of specific grown-up devices. I then subscribed to certain specific channels with kid-friendly content and I use GrayJay to pull the content from those channels onto the NAS and let Plex host the videos. We periodically update the subscription list in GrayJay and keep the offline copies available for them. Benefits: - We the parents very specifically curate the content. - We the parents mare actively involved in the content they watch - We have offline content ready to go for trips or Internet outages. Cons: - It takes a lot of parents time to curate content, but subscriptions to good channels does help. - GrayJay download --> export is manual, which is also very time consuming. I am looking at ways to automate that process. I gave up on YouTube Kids a long time ago. It's a neutered YouTube client with broken search and useless filter options. The main YT client is safer at this point.

@Stephen Chr Agreed. I was recently on the interview team for a candidate that our company was hiring and the "portfolio ask" was to pick any tech project that utilizes the things in the job description and demonstrate as much of the the planning, research, discovery and implementation phases as possible within a week. The idea is that the candidate picks something _THEY_ want to do and we don't give any guidance on what they should pick. The first of two rounds of interviews was the standard "Get to know you" stuff. The last was "Pitch your project idea, show how far you got and describe your journey". Nobody is expected to have a working prototype completed but having _something_ down that shows us you _actually_ know what you're doing outside of some slips of certified paper is invaluable. The ironic observation: The more certifications a candidate has, the less likely they were to complete the assignment. I don't know if it was arrogance that their bits of paper carry them further or if they bits of paper didn't mean what the titles had written on them.

@Stephen Chr Exactly. Our ask was specifically a project the candidate was personally interested in and not related to anything the company does.

@Ken Pryor Same. My wife is usually okay with things if I'm sensible and seek out a decent deal... so I tend to go with used hardware or smuggle when necessary. I keep purchases somewhat clumped together but clumps are spread out over time to soften any blows. Patience is key.