I probably go through at least one full discharge cycle a month, if not more because the power around here suuucks. (The NAS goes down, but I leave the network gear up until the UPS dies, because fuck it, why not.)
It's also a ~10 year old UPS that likes to eat a $25 battery every 18 or so months so I just haven't really had any justification to replace the whole thing yet since there's an awful lot of $25 batteries in a new UPS.
I replace the batteries in my UPS every 18 months, and don't try to outlast power outages.
I have everything configured to shut down if the power goes down and stays down more than 5 minutes, which is ~20% of the maximum rated runtime. (I'm using repurposed desktop hardware that loves it's watts as a home server.)
I picked the low number for the reasons you've outlined: even if the battery is severely degraded, it's probably not THAT severely degraded and it's a safe time span to ride out short hiccups, but still well under the runtime limits so that a safe shutdown can happen.
That and I've noticed that, typically, if the power is down for 5 minutes it's going to be down for way longer than 5 minutes, so it doesn't matter and I'm not going to have enough batteries to outlast the outage.
I've seen people stuff GPUs in 1 and 2u cases so you probably could depending on how deep the case is/how the internal layout is and if you can find somewhere safe to stuff it.
I dunno, I know a bunch of "senior" administrators that would be this blindingly stupid.
They just assume they're smarter than everyone and thus don't need to do anything to hide their shit because nobody could possibly figure out what they're doing.
So far I've yet to meet one that's actually right about that.
I mean this with all the love in the world, but you're surprised that mods, in a gaming subreddit, are complete assholes?
This last year reddit has just been fucking trash all around, and I just... quit going, even though I still have things that the Fediverse just plain can't provide any useful content for.
Not because of anything spez did, or whatever stupid-ass money seeking bullshit they're trying next, but just that the content is crap and has gone from interesting and useful to trash. For example, the selfhosted subreddits went from cool software to being stuffed full of people who do NOTHING but argue about hosting email, and marketers telling you how easy it is to host something you can use to spam people. Just a degradation from good to shit.
And, even when it's not immediately obviously worse, it's still full of shitty bots, and the mods run around like they're fucking royalty and we should lick their boots.
Seems like it's not worth the tens or hundreds of thousands you're going to spend fighting in US Federal court over the next 5 years, but also not my money.
I just went with a plain boring Ubuntu box, because all the "purpose built" options come with compromises.
Granted, this is about as hard-mode as this can get, but on the other hand I have 100% perfect support for any damn thing I feel like using, regardless of the state of support of whatever more specialized OS is for aforementioned thing.
I probably wouldn't recommend this if you're NOT very well versed in Linux sysadmin stuff, and probably wouldn't recommended it to anyone who doesn't have any interest in sometimes having to fix a broken thing, but I'm 3 LTS upgrades, two hardware swaps, and a full drive replacement, and most of a decade into this build and it does exactly what I want, 95% of the time.
I would say, though, that containerizing EVERYTHING is the way to go. Worst case, you blow up a single service and the other dozen (or two, or three....) keep right on running like you did absolutely nothing. I can't imagine maintaining the 70ish containers I'm running without them actually being containers and/or without me being a complete nutcase that runs around the house half naked muttering about the horrors of updates.
I'm not anti-Cloudflare, so I use a mix of tunnels, their normal proxy, as well as some rawdogging of services with direct port forwards/a local nginx reverse proxy.
Different services, different needs, different access methods.
After a major illness, hospitalization, and two upcoming surgeries for said issues, I fully support any use of AI that can be used to bludgeon health insurance fucks.
Sure, but isn't the mixture of rust and c based device drivers and kernel ABIs kinda the whole kerfuffle that's going on right now? Seems like the rust faction would probably NOT want to grab some duct tape and stuff all this back together, but slightly differently this time.
I think we're kind of saying the same thing: making something that boots as an MVP isn't the most difficult thing but is still a much different and simpler project than making a replacement kernel for Linux.
If you really wanted to be a legitimate actual replaces-Linux-for-all-the-things-you-use-Linux-for kernel, you'd be biting off years and years and years of work on drivers. I mean, just look how long it took Noveau to go from 'kinda works' to 'actually viable' and that's just a subset of GPUs from one vendor.
I'd also add that if you cared about server hardware, you've got a much larger driver footprint with a lot more weird behavior shit than on desktops which are, honestly, just a couple dozen combinations of chipset, sound, ethernet, and usb controller in any given generation.
And sure, you could lean on the work that's already been done at this point and probably do it faster, but it's still a massive undertaking.
The problem is these stupid fuck zero-tolerance laws and policies. I'm not nearly in a position to argue if they are or aren't intentionally racist, but they're certainly more likely to impact non-white students.
But, still, the shit I did in middle and high school would have resulted in me being kicked the fuck out if it happened now, so I kinda wonder if we're doing very much the wrong thing by ratcheting up the punishment for shit that, in the past, would have been more of a time-out and less of a life-wrecking moment.
remember 20 years ago when performance really, really mattered
At this point, you could almost just assume next gen CPU is 10% faster than the one that was 10% faster last year.
GPUs are a little more wild, but that's more a side effect of all 3 having wild swings due to hardware changes and driver updates but it's still just a single graph you have to update every few months. Not really enough to run a business on.
Any article I've seen from them has either been so low detail it feels like someone used ChatGPT 0.1 to write it, or it's so blatantly just and advertisement for whatever sponsor they're talking about.
Of course, that means AnandTech probably was still decent and since it was decent it cost money to run, so fuck it, let's just see how much we can get nvidia to pay us to post fluff pieces instead.
I'm not sure I'd call device drivers 'finishing work'.
The MAJORITY of the kernel is that pesky bit of 'finishing work'; there's a fuckton of effort in writing drivers and support for damn near ever architecture and piece of hardware made in the last 30 years.
You could argue you don't NEED to support that much stuff, and I'd probably agree, but let's at least be honest that the device driver work is likely to be 90% of the work and 90% of the user complaints if something doesn't work.
Little bit of A, little bit of B.
I probably go through at least one full discharge cycle a month, if not more because the power around here suuucks. (The NAS goes down, but I leave the network gear up until the UPS dies, because fuck it, why not.)
It's also a ~10 year old UPS that likes to eat a $25 battery every 18 or so months so I just haven't really had any justification to replace the whole thing yet since there's an awful lot of $25 batteries in a new UPS.