I’ve been using the glinet comet kvm for my homelab and have no complaints. Their cloud is optional and I don’t use it. The built in tailscale client does what I need it to. I use it with their ATX power accessory to manage physical power on/off when needed.
Given that these things have bare metal access, keeping them off of the public internet seems wise no matter what though.
Keeping these kind of management devices off the Internet seems prudent. But how do you do that and still get Tailscale to work? Assign the device to a separate vlan that is restricted to only talk to Tailscale? Otherwise, if the device is on your regular network, it will still be connected to the internet.
Untrusted devices can sit on a separate VLAN or get WAN blocked, you can still reach them internally, and from any other device on Tailscale. You just need to expose the subnet via Tailscale subnet routing.
Yes that is how you arrange how the device can be reached through Tailscale.
What I was wondering was: In order to get the device to talk to Tailscale to be able to reach it you need to give it access to the internet to reach Tailscale. But now I understand your answer and it is to let the device sit somewhere in an enclosed network and then through another trusted Tailscale node route any traffic to it using subnet routing. Thanks!
I came here to say almost the same thing - except I don't think there's more space between the puzzles. From the post:
Why did the number of days per event change? It takes a ton of my free time every year to run Advent of Code, and building the puzzles accounts for the majority of that time. After keeping a consistent schedule for ten years(!), I needed a change. The puzzles still start on December 1st so that the day numbers make sense (Day 1 = Dec 1), and puzzles come out every day (ending mid-December).
Honestly, I prefer this. I've never had time to finish AoC because as we get later into December, I'm crunched with wrapping up work things so that I can do more family things and prep for the holidays. I look forward to this every year, and every year, I can see that I stop around day 11 +/- 1 day (except for 2021 when I somehow got to day 17).
I came here to say almost the same thing - except I don't think there's more space between the puzzles. From the post:
Why did the number of days per event change? It takes a ton of my free time every year to run Advent of Code, and building the puzzles accounts for the majority of that time. After keeping a consistent schedule for ten years(!), I needed a change. The puzzles still start on December 1st so that the day numbers make sense (Day 1 = Dec 1), and puzzles come out every day (ending mid-December).
I tried a Mac Mini for a while, but they're just not designed to run headless and I ultimately abandoned it because I wanted it to either work or be fixable remotely. Issues I had:
- External enclosure disconnected frequently (this is more of an issue with the enclosure and its chipset, but I bought a reputable one)
- Many services can't start without being logged in
- If you want to use FileVault, you'll have to input your password when you reboot
Little things went wrong too frequently that needed an attended fix.
If you go off the shelf, I recommend Synology, but make sure you get an Intel model with QSV if you plan to transcode video. You can also install Synology OS to your own hardware using Xpenology - its surprisingly stable, moreso than the mac mini was for me.
> Get a Synology Intel model with QSV if you plan to transcode video
How about if you _don't_ plan to transcode video? For example, this years models DS425+ uses the (six year old) Intel Celeron J4125, while the DS925+ uses the (seven year old) AMD Ryzen Embedded V1500B. Why choose one over the other?
That’s a damn good question. I’d probably go with the intel model that had qsv just in case. Especially compared to an equally ancient AMD processor that doesn’t.
I agree with you in principle, but 30 years is probably the wrong number. I have a house in coastal Maine, built in 1997. It's coming up on 30 years. I assure you, it's vastly different than the my first house (1941) or my last (1953), in good ways.
But to your point, we consider way too much to be "historic" and I'd like for that to change. You really should be able to tear down almost anything you'd like and rebuild as long as it's to code/zoning, and zoning needs to be cut back to things like dimensions and use, not appearance. Being old shouldn't make something eligible for historic preservation on its own.
Code should be about fire safety for the fire department if it burns. if you can't build on the lot then it needs to be because of something there - or if they are planning to build it (a new utility) then they need to rebuy the easement every 10 years until they do. Height is limited only by airport flight lines.
everything else is none of your business. (Okay, I might have missed something but it is on those lines)
Height (number of units, really) is limited mainly by the bandwidth of the roads, water mains, and sewers in the area. These are things that are expensive, very expensive to expand. Of course in practice height is firstly limited by zoning to protect existing properties' owners, but you can't ignore the infrastructure needed for larger buildings.
In the United States something simply being old in no way makes it historic in any legal sense (e.g., contributing to a local historic district): integrity[1] matters if you are trying to legally deem a resource historic and worthy of some sort of preservation effort. Generally speaking buildings, structures, or objects need to be at least 50 years old, integrity aside (but there are exceptions if they are particularly noteworthy).
I tend to disagree with the need to tear stuff down just because it's old: it's so terribly wasteful. We need to get better at adapting, reusing, and adding on to older buildings. Granted, when developers just want to use the cheapest materials possible and build something that will start to have serious problems in 20 years, it's a problem, to say nothing of the loss of serious knowledge in various skilled trades.
[1] In most cases the National Register of Historic Places aspects of integrity are used for evaluation: integrity of location, design, setting, materials, workmanship, feeling, and association.
Great question. The unit is pre charged for 25 feet of lineset . I was right at the limit for my install.
I installed the lines, did all the vacuum related work. Then just cracked the valve on the unit to distribute the refrigerant.
I had a bottle of refrigerant on standby from a buddy. Didn’t need to use it . I was going to get the EPA cert, if my buddy didn’t exist. I heard it’s super straightforward
The vacuum work, finding/fixing leaks, etc can be a real nightmare, even with the expensive equipment. Personally I would much rather pay an HVAC technician to just come and pull the vacuum. They already have the gear, and they can diagnose when your vacuum still doesn't pull all the way for the 3rd time. Then if you don't get a precharged unit, they can also fill it.
that is IF the technician cares enough... unless you guys have stringent regulatory/right incentives in place. sometimes tech just relies on the brand reputation and pull for 10mins calls it a day
Given that these things have bare metal access, keeping them off of the public internet seems wise no matter what though.
reply