 Welcome to the homelab show, episode 88, Q and AI. I think it's a fun way to put it because we are going to bring up after we do the Q and A first because we're going to give time to the humans that have interacted with us first. I think that's what we decided to do here. Well, we start taking questions from AI, or if we get AI questions in, I think that's going to be a different problem. Yeah. That's a whole different problem. AI starts joining the live stream and asking questions. We'll deal with that problem when it arises. Someone's going to connect it to the chat. I feel it happening. We also have two ways that people give us feedback. It's feedback at the homelab.show. We've been trying to promote that to make it easier for you to interact with us and send us emails. Please frame them into formal questions like you're playing Jeopardy. Because sometimes some people wrote a lot and we're trying to decipher it's a question or not. But hey, we still like hearing reviews. Sometimes just sharing a short experience when things is fine as well. Also, there is a form that you can fill out that goes to a doc that is still available on the site for now. We're trying to decide which is the best way to handle the feedback on the show. But one of the less, we love hearing from you. What's the first question we're going to start with, Jay? Yep. We're going to get into Q and A. Let's do it. In no particular order because I'm not as organized as I want to be today. One person asked, Michael asked us in reference to episode 80 where I'm talking about syspref because I used to actually be a Windows admin before getting into Linux. So I have some knowledge there. So I mentioned syspref. And then Michael asked if there's something that you have to do for Linux machines to generalize them before taking an image much in the same way you do with Windows. And the answer is yes, but it's not as required. It's not as demanding. Whereas Windows systems, they have the registry. They have all these different things. And if you try to clone a Windows install from one install to another, it might work fine, but it may not. That's what syspref helps you take care of. But on Linux, you could just do nothing. And it would work just fine, although you would have some side effects, like perhaps the same IP address being assigned or host key mismatches and things like that. So there's a few things that you should do. And one of those is to reset your SSH host keys. I like to delete the command history on any user that's in the image. You don't have to do that. I also like to empty log files. You don't have to do that. But I would say the least you should do is generalize your SSH keys or the host keys. Because if you don't do that, then every machine you set up is going to give you that snarky message from SSH, which is actually normally a good thing if you see this that it's warning you about it. But it'll be a false positive and tell you that there could be a man in the middle attack. And no, there's not. It's just the same host keys are on every machine because they're in the image. So we covered this in more detail in a previous episode. And there's also a video on my channel that will go over this. I think there might even be a few that covers this. So the answer is yes, but you don't have to do as much. So I guess that's where I'll leave it. Yeah, it's a whole lot simpler cloning Linux. The lack of the licensing also makes it easier because it gets further complicated with Windows and license activation based on hardware changes, etc. So. Yep, it sure does. See, licenses just complicate all these things. It's a weird way they do it. They're legal stuff, right? It's obviously legal stuff. Legal stuff. Lawyers, just keep them out of the home lab. Keep them out of your lab. Unless you are a lawyer, then you can still work on your home lab and still be a lawyer. I'll cover line 170 here. The Tom and Jay really enjoy the show. It would be great to get some Kubernetes content, a couple of suggestions. And they dive into some really bigger topics like Longhorn and Cilium and Grafana and running K8s. Me and Jay aren't in-depth Kubernetes people, but Techno Tim, our friend, is. He does this for a living. He's covered like every topic this person had mentioned in the question is actually something that Tim already has a video on and is covered in-depth. And he's well, I would say, substantially more well-versed than me. Was he a little more well-versed than you? I mean, you do more Kubernetes than I do. I don't know where they're at. I did ask my question. I don't know. He's very talented and he knows a lot. That's what I do know. I do cover Kubernetes on my channel. And right now, there's like a soft pause on my coverage of it because I'm currently in OpenStack LAN because I have that series uploading. But sometime after that, I'm going to start diving into more Kubernetes content on my channel, but you don't have to wait for me to do it because Techno Tim has content on it right now. So if you have an appetite for it right now and you don't want to wait for me to do it, then absolutely check out his. And even if I do have videos on my channel, watch both channels. Why even watch this one? Watch them all. And one of the things me and Jay have been working on in building what we will call our team of friends. They're trusted other YouTubers that we can definitively tell you because they work in the industry. They're industry professionals. They aren't just YouTubers. They're someone who actually does this for a living. So their knowledge on it is very concise. They're not just shooting from the hip. They didn't just read the book yesterday and make a YouTube video today. Techno Tim actually has a job doing these things in building automation systems. So it's not just a hobby, which is why his videos are so accurate. I've certainly watched a few of them because he's got some really solid intel on there. So when we say watch him, we don't mean it like maybe he's okay or good at it. We actually will vet him to say, yes, he's really talented and has great information on there. So yeah. And just some other random recommendations. Veronica explains is a great one to check out. Jeff Geerling, I don't even know if I need to say that. Is there anyone? Everyone knows. Craft computing and there's others. Maybe one day we're actually just going to come up with a list of YouTube channels. But if you look at our YouTube channels and look at the feature channels or whatever, you'll get a list from right there. We might come up with a more formal place, but for right now we have that on there. Yeah, we're working on the back end to bring you eventually we want to put ourselves all in one page. So you can say, hey, these are all, because we do all talk. And we're all like a same group of friends that all care about teaching you some of the best stuff. And I see someone through level one text. Yes, WINDL is absolutely a friend of the chain. Oh my God, how did I forget? See, this is the problem. Eventually you'll end up on a website with a list. Yeah, that's why we need a list because I'm always going to forget somebody and then feel bad about it. Let's see, what's the next one we have here? Oh, I think that might have been something about something I mentioned here. Let me, I've always turned on rapping because the questions always go off the page. So let me fix that here. Okay, so this one is in regards to the of Home Assistant and the tool that I mentioned regarding integrating it with a Windows system because, you know, no judgment. You might have a Windows install. Doesn't matter. Everyone's in charge of what software they use in their own Home Lab. And if you do use Home Assistant, which I also use in Windows, then there's absolutely something to look into there. I'm just trying to remember the name of it. And I think what I'll do is just maybe during another question, I'll Google it and just try to remember. Because I don't remember the name being very memorable, either that or I'm just forgetful. But I'll definitely make sure that the answer to that question exists before the end of the podcast. But definitely something that I would recommend. And I did list, you sent it to me last week, and I made sure to add it to the show notes for last week. Oh, you did. Yeah, it is in, it is in the show notes of the Home, of that Home Lab show where you talked about Home Assistant. So that's definitely in there. Okay. So, yeah, there you go. So check out that one. That was, was that our previous episode, I think? Yeah, last episode. Yep. Yep. Okay. So check, check out our previous episode and yeah, I'll have the answer there. And basically from what I remember, it's a service that runs kind of like any other Windows service. And it allows communication between Home Assistant and Windows. And there's a lot of things that you could do with it. I never really kind of looked into it very deeply other than a one-off thing that I needed to do. But I think somebody who has, you know, more of a need for it would probably do a lot more with it than I've ever done. Yeah. The next one is going to be the VLAN one. Should we talk about that one? That's a, that's a big question. Yeah. You know, this is something that I think I, I ran into at one point early on. And I wonder if this is, there's certain, certain growing pains. I don't know if I should call them growing pains, but maybe just like milestones, right? When you're, when you first get into Home Lab and you start with one server, then you end up with a entire data center. But one of those is trying to figure out how to create your network. And it's one of those things you want to do if you can at the very beginning and have the design, you know, already created and just go along with that. And you'll never have to go back and fix anything. The problem is most people are learning and that's why they have a Home Lab. So they're, they might not know how to carve up their IPs or networks and things that have a good idea. But the actual question is when is, when is it the case that you have too many VLANs? Is it less, is it more, should you have a bunch? And I mean, what's the right number here? And my answer is not much of an answer to be honest, but it's how I feel. Each VLAN has to exist for a reason. When you first get started with VLANs, it's like a superpower. Like if it's something that you didn't understand, but then that one day comes and you're like, oh my gosh, I got it. I totally get it now and it just clicks in your mind. Then it's just that excitement around understanding something. You just want to like install it on everything or use it everywhere. Next thing you know, you have 200 VLANs. But the actual thing is each VLAN should be, should have a purpose. So every time you create a VLAN, don't just create them for the sake of creating them. Create them if you have a purpose. And sometimes security is the best way to go because you just base it around security. You want to keep things segregated and that's a great place to start. And don't go too far with it. I've covered this before in my PF sense setup for home. I've had a lot of feedback, if you will, where people say you didn't create 30 VLANs like I did. And the problem is those are sometimes the people who end up posting all over the forums going, how do I manage all the VLANs? And by the way, nothing talks to me. I put every device on its own network, but I need some of these devices to talk to each other. And then they spend a lot of time figuring out that once you start setting up separate subnets, and that's really what you're doing here, some devices may not, depending on the device, and this varies in those famous words of it depends, do those devices allow communication, no matter what firewall rules. And sometimes there's helper apps to get the data across there. But for the most part, stick all your IoT on one thing. And have maybe more trusted devices on another, and maybe your most trusted devices in yet another security network. Now this can be very important because the thing that we preach about here on the Home Lab actually happened at corporate level scale just a few days ago, which was a zero day in a product. So as we talk about when you publicly expose something, what if there's a zero day? And what if came true for some very large companies, and that's because a product that was public facing that sits on a lot of companies' networks. Now, I know two tales here I can tell you, a company that survived, and a company that just had a big, large disclosure because they're publicly traded because they didn't quite, they survived it, but it's a much bigger mess. What happens is when one of these public-facing devices has breached, where is that lateral movement? That's what you're usually trying to protect from when you segregate your network is lateral movement. So devices reaching out, and what if someone poisoned where that device reaches out to so something can get back in, and it's a rare, but could happen, but then where could it go? Like, let's talk about security cameras. Where can it go? Well, I don't give the opportunity for the security cameras to reach out, but maybe you used a UV camera that needs internet, and it does reach out. And then somehow it laterally can move side to side. Well, don't put it next to your computer, but then again, also keep a firewall on your computer as good measure, because that's a good thing to do. It all comes down to what you need to segment and how you want to prevent some of that potential lateral movement from happening. But don't go overboard with it. That's our best advice on it. You need as many separate networks with security as you need, as you can handle, and it doesn't aggravate other users that may share the network with you. I'll give a perfect example of why this is a great idea, because you always think that one of the things that we mentioned on this podcast and what's mentioned on the Enterprise Linux Security podcast, you know, it's just, oh, that sucks that it happened to that person. That must have been a bad day. But I don't know if people think that something could happen to them. So when we bring it back to VLANs, you want to separate things. And I had this situation, and I was going to bring this up in the review for the Minecraft Mark II device that is now not going to happen because I'm not going to be reviewing it for obvious reasons, but I do have one. And at one point, I was just sitting down editing video and I heard background noise and I'm like, what is going on? I heard this super loud fan. I'm like, everything else is off. And I look over, it's the actual Minecraft unit, the Mark II fan's going crazy. And then on the display comes this FBI warning that the computer has been seized and locked and all that. And I knew exactly what that was the minute that I saw it. And that would have been really bad because I mean, it is bad, but I have all of my IoT devices on a dedicated IoT VLAN, including the Mark II. And there's no rules to allow anything there to talk to anything else. So even though that sucks, nothing else happened. That device, theoretically, if it was built to do that, could have contacted other devices on my network and started crawling around. And next thing you know, everything is probably crypto-lockered, but nothing happened except for that one thing because it's on that VLAN. It can't get to anything else. It can only get out to the internet. So I saw it happen, pulled the network cable, wiped everything, which I shouldn't have done because I probably would have wanted to explore exactly how that worked. But the point is, if I didn't have that VLAN, that would have been a bigger problem. And I think that's the best way to think about this is, what's the worst thing that could happen? If someone breaks into your main computer, what else would they be able to access? And some things you need to access, like your printer, for example, you want to print, don't you? So you have to be able to access that from your computer. But other things, maybe they just need to go out to the internet and be segregated in some way. That's the way to think about it. How do you want to carve that up? How do you want to separate things to minimize your workload, not create a bigger support service, but just make things more stable and secure? That's the goal. Right. And the question of how do you decide what's IoT? Well, that's a little bit fuzzy, because I often have told people that your phone is IoT and it makes people, those people into an immediate panic. But my phone is my precious. And I'm like, no, I'm sorry. Your phone is, you know, and most likely Android or iPhone. And you probably wanted to talk to your, let's say, Chromecast, because that is literally my use case. I have a Chromecast attached to my TV. I have my phone on that same network. I have MB on that same network. Why? Because they are all IoT devices to me. Yes, that does include MB that runs on my NAS that talks to it because I would like MB to talk to my Chromecast and I like to control that via my phone. That is all one big happy IoT family because it works well together. Trying to separate out your phone. What attack are you trying to? Your phone is meant to be, so to speak, in a hostile environment where, yeah, there's the potential that someone's going to take over my Chromecast. But who's going to really take over my Chromecast? Is Google going to send commands down from the cloud and scan my network for my phone? That they also happen to own and then try to do it? At some point, you're getting to levels of ridiculousness and that's not where your threat models are coming from. They're just not as prevalent there, especially when you have trusted brands. Trust as in, I don't give them privacy trust. I trust them with security. So let's talk about what type of trust we're saying here. I trust that Google is not good for privacy. I think they're good at security because they don't like anyone else but them having all my private data. So just think about how those threats are before you define it. And if you are using random things you found on Amazon, it has a weird name. That is the scary IoT stuff that has weird firmware on it. That can go on a completely separate network unless it needs to. I generally don't even want toasters or refrigerators on my network. But if you feel insistent that your washing machine and toaster need to be on the network, maybe you do want an appliance network. So those can be considered IoT. Then you could have a situation where the fridge, which is the thing that you put your spam in, ends up sending you spam. But that's not a situation you want to be in the middle of. As far as where to draw the separation between IoT and not IoT, that's a whole other topic in and of itself. But what I will say, possible thought experiments there, if you have to jailbreak it to SSH into it, it's IoT. Okay. You were the one that installed the operating system yourself. It probably isn't, but it still could be if you're the kind of person that builds your own IoT devices because we do that kind of thing. But if you take all those edge cases away from it, if you have to jailbreak it to SSH into it, it's IoT. Plain and simple. If you haven't or weren't able to install the operating system yourself if you wanted to and you had to jailbreak it just to put a different image on it, that's also IoT. And I think that probably covers what? 90% of the edge cases, I think? And I have printers, and we do this for businesses as well, putting printers on the same network as the computers because I don't hate myself. And the statistical likelihood that a printer would become the attack point is not zero-likeliness, but hopefully, one, don't open a printer to the outside world so no one should be accessing it besides local computers. And if someone were to take over the printer, it means somehow they got on a local computer anyways to try to take over the printer. Well, the printer's less powerful than the local computers. That's why most attacks start from a breach in the computer. The threat actors aren't as likely to own. Hold on, I got this really powerful, nice i7 workstation with 32 gigs of RAM. Let me wander over to this really low-powered printer to launch my attacks from. If they're on your network, they're going to launch attacks from the device they compromise. It's easier to control. So there's a few thoughts around that. So one thought about printers. I don't want them to be anything to be cloud-enabled if you can help it. But there's a way you can use that to your advantage, though. If you buy a printer that is cloud-enabled, then what you can do is literally segregate that thing, put it on a different VLAN, do not let any computer talk to it because it could be accessed from the internet. So make your printer go out to the internet, allow that. Don't allow it to talk to anything else. And if your computer has to go out to the internet and then back in to print to the printer, you're not going to notice a delay. Obviously, a company with a lot of people probably wouldn't want to do that. But for home users, if it's cloud-enabled, great. You could print from the internet. Print from the internet. Because if there's a firewall rule in between your printer and your computer, your printer's going to think that there's nothing else on the network. Your computer's going to think there's no local printer. That's fine. They'll go out to the internet and meet each other there. And then it'll be more secure. But then again, that's another rabbit hole. Everything is cloud-enabled nowadays, and I'm not sure I like that. Yeah. And I'm positive HP is working on an even worse way to implement this. If there's something they've become skilled at as making worse and worse printers, I've, yeah. I missed the early days of HP when they were a good company. I mean, like 20 years ago, like I still have one of those printers for 20 years ago and it still works because, you know, ah, nonetheless. Yeah. One of them I'll quickly address is someone asked a question. They have a bunch of domains. And they said, can I monetize these as in, can I use all my HomeLab stuff to start like some type of hosting because they have a fast connection and a lot of stuff in their HomeLab. That's off topic a little bit from HomeLab. But from just a quick concept, I'll at least address it. It's a lot more complicated than you think from the business side of the liability you may take on by hosting things. What if you suffer a breach? What could you be sued? How would you be held liable? It's not about that. Could you do it? Yeah, sure. You can spin up some of that storage you have. But you're also compared to hosting in a data center. You're not going to offer or be able to offer, I should say, the redundancy, the multi-location for the data, and the resiliency of what if something happens at your single location. So that was an odd question. But I think it's interesting to think about because the HomeLab is not necessarily unless you're working towards a job, a moneymaker for you. But it's also not easy to turn into a moneymaker. Maybe hosting some gaming servers or something might be more profitable amongst friends where you have a friendly agreement. But yeah, to do it is going to be a little bit more complicated than just the technology side of it. It's the resiliency. It's the legal problems you may have. And on top of all that, you've got to market it. Get people to come to the service and use your service over cheap services already out there. Because right now, storage is generally speaking in a lot of cloud services and they're not that expensive. So can you compete or make money at that level? Usually it's the management is where you can make money. So if you use those skills you learn in HomeLab to get even a part-time job to help feed your HomeLab and justify buying more hardware to test on, that justifies some job skills that brings you back to the whole cycle there. The other thing about this too that you have to... I'm not encouraging anyone to not go into business. I'm just letting you know one of the things you will run into if you try to do that. And again, do it if you want to. And I would love to do it myself. You just have to be under... You have to know the fact that you are going to deal with some anxiety, some stress you didn't expect. And what I mean in particular is when you have somebody who's asking you to provide services for them, they're asking you to do that because they don't know how to do it. If they did, they would do it themselves. They're asking you to do it because they want someone to do it that knows how to do it. And their understanding of how these things work is going to be severely different than reality to the point where you're constantly correcting people. And the best example of this is I remember a client who actually was upset because they wanted two sites connected but without internet between them, without VPN between them. And I'm thinking, how do you want me to accomplish that by a least line or something? But they feel or felt that there was a way to still do it. No matter how many times I told them it doesn't really work that way, it was quite a frustrating experience of things like that happen all the time. So it's fine if you know that's going to happen but just keep that in mind. These people don't know how to do this stuff that we do. So we will be explaining quite a bit of things to them if we go into that business. Yeah. Question, and I think this is the last question we have, a question for you, Jay, because you were trying to think of the name before the show. Is it Tarsnap? Yes. Okay. Now we can answer that question and we know the answer. I'm giving the answer. We're going to start the question now. Oh, right. So you're referring to the one that has, the individual has BSD on his network. Yeah. So it was one of the longer messages, but my understanding of the problem here is that Proxmox backup system, according to Andrew who wrote in the question, doesn't support free BSD. And I'll just take his word for it because I never looked into that. I personally don't use BSD. So people that do use it will know more than me, obviously. But if that's true that it doesn't work with BSD, then Tarsnap might be a great way to go for that. I think that's something that BSD people, last I looked anyway, really do enjoy. And if I remember correctly, it's not specific to free BSD or anything. I think it works on Linux as well. Again, double-checking on this has been like probably four years since I looked into this, but it was something that came very highly recommended. And isn't it true that George Lucas, Michael Lucas? Michael Lucas. George Lucas is a child. Yeah. No, he wrote a book on that, right? Yeah. I believe he has a book on it. By the way, if you haven't read Michael Lucas' books, please go ahead and do so. We've recommended him numerous times because he's got the definitive guide on a lot of things like SSH and all kinds of fun stuff and technology. He's a good writer for all that. It's called Tarsnap Mastery. Yes. Tarsnap Mastery. He's got a theme going. Got a theme going. Check out that book. His SSH Mastery book is so far my personal favorite. It tells you things about SSH. You never knew what to do. The Tarsnap is definitely one of the things. I'm looking at the book right here. So if that's something that helps, I hope it helps. Yes. But we should reach out to Michael Lucas to be a guest. I think so. We talk about him all the time and nobody knows. A lot of people do know, but I know there's some people that don't. So probably should get him on. Yeah. Does that conclude all the questions? I think we got him on. I think it does for the most part, at least. But yeah, I think that pretty much covers it. All right. Now, let's pivot over to chat GPT. And why is it in a Q&A show? Well, honestly, I think it's a great place to ask questions. But I know it's not 100%. So I'm not saying to trust it, but where I do find the middle ground. So I've been playing with it more and more, and I haven't done a video on it because I'm annoyed by things that are overhyped. If it's overhyped, it's probably just overhyped. It's companies trying to make money on it, et cetera, et cetera. And my understanding, the launch of being had an epic failure of just straight up giving wrong information. I had chat GPT write me a biography about myself, which is easily fact checked with a quick Google search. And Google was right and chat GPT is wrong. So we're going to say it gets a lot of things wrong. But there is a lot of things that gets right. And one of them that I found really interesting as a use case is writing frameworks for code. Now I say frameworks, but it'll actually write some code and then explain what the code does to you. So chat GPT from a code Q&A, like, hey, I need a bash script to do this, this, and this. And it will create annotated code. I mean, like it puts nice comments in it to explain what each section does. You can also work with it to keep iterating till you understand the code and maybe ask what this section of code does. The fact that you can go back and forth with that which chat GPT makes it really interesting from a home lab standpoint, especially when you're getting started. Me and Jay talk about using Ansible or bash scripts and things like that. But when you are just getting started with a home lab, this is a blank sheet of paper that you're staring at going, man, what's the first thing I type? What's the first couple? What is a she bang? Why would I put that there? And those are questions it seems to know very well. It actually does programming well because when it comes to things that are extremely deterministic and not really opinion, like code that works, it seems to understand it very well. I have a video linked down below from a channel. Many of you probably follow Dave's Garage. Dave's a longtime programmer for Microsoft. And I thought he did just a great job of explaining how to use chat GPT to write code. He has it doing a few different things. He even talks about how he wrote the code. Then he compared to what chat GPT wrote based on the concept he asked. And then he translated it into different languages, which is actually kind of fun too. Because you can say, write this in Python, write this in Bash. I think it's just one of those things that I really thought about it when I started using it because I started writing some PowerShell that I'm not very good at, but chat GPT wrote a PowerShell for this. Then you can say write a menu. Then I wrote some Bash scripts. I said, you know, and make this Bash script compatible for the thing it has to do. I want this package to install, but I want to check first if it's a Buntu-based install or a Debian-based install. And it made all the different parameters in there. You can even have it, you know, generate some menus and things like that. This is a great framework for you to start putting it together and not have that really how do I start type feeling. I thought that was cool because it really gave me the base I wanted. And, you know, if you're old hat at this, sure, you're going, why would I use this? But even watch Dave's video and he'll even show you, you know, a guy who is a long-time coder, knows many languages very fluently, was still impressed and thought of all the great ways to use it. I think his video is not hyped at all. It is very concise and walks you through code examples and it was kind of fun. Yeah, that sounds fun. And I haven't used chat GPT yet. I've been, you know, very, very deep in video production. But, you know, hearing you talk about it, I'm kind of tempted to just ask it to develop a proper sequel to Chrono Trigger. Just see what happens. Yeah. Because we need that sequel. Come on. Chrono Trigger is good, but it wasn't Chrono Trigger. Maybe chat GPT can help. Just don't ask it to create a self-aware AI bot that, you know, named Jarvis that'll take over the world. Now, don't do that. But yeah, I don't know. Of course I joke. But, you know, there's something to be said about looking at code examples. And if that helps, I think that might be a good tool for that. Yeah. You can ask a question, explain it. You can help with the regex. And I don't do regex often enough. So I actually have a bunch of little things I need chat GPT to do. And once I have a lot of these done, I might make a video on it. But I just have, I know what I want extracted, but I know there's going to be a lot of parameters because the way this certain log comes through. So I'm like, you know, I'm going to feed it those logs and say this is what I want out of these logs is this particular thing and see how it writes the regex for it or some type of parser for it. And I have a couple tasks like that. And then of course, maybe I'll post them all on GitHub and it'll be kind of handy. But I just really thought this was worthwhile for the homeland people, especially because it's free. There are times when you get the high usage stuff, they do have a subscription now if you want to avoid that. But for the most part, keep trying, you'll hit it for free. And I think it's a good place to get started. Now, where you shouldn't use chat GPT, I will go back to this. I asked it some questions like let's compare Synology versus TrueNAS. And it got some things wrong right away. I did try to ask it to do some insecure things. Because you can sometimes coax it into explaining a bad topic badly. Like, hey, tell me why it's a great idea to open up Windows RDP. I was actually shocked. I played around with it and I was unable to. It kept telling me, no, it's not a good idea. It'll argue with me. Oh, that's good, that's good. Yeah. Well, I tried to ask it. Tell me why it's a good idea. Sometimes you can try and coax it in this stuff. You got to remember, it can be very biased because you could ask it a question that you're leaning into a bias and it can give you bias confirmation because you asked it to say, tell me why this is great. Or if you started, tell me why this is bad. So those things are where chat GPT will lead you off the path and also it can be inaccurate. But from that coding standpoint, just bringing it back to that coding stuff, I think it does a really killer job on that. That is, I think for the Home Lab, I want to play with it and see how it writes Ansible and things like that. Can it write Ansible playbooks? Create a playbook that does these things and will it create the base for that? And once you have the base, you can go, oh, I think I understand the structure and start iterating forward on it. I think it's just a handy tool that's free to use in the Home Lab and start building your automations with it. Yeah, but if it's going to lead us down a different path, that would mean it's manipulating us and if AI is manipulating us, that's a really bad sign. No. Chat GPT has a lot of fans and I think it's even going to be, if I'm not mistaken, integrated into Opera at some point. I'm not sure how I feel about that because I think the option should be there for everyone that wants it. As long as it's not stealing CPU cycles when it's idle or something, that's fine, but apparently Opera thinks enough of it to put it in the browser. And I asked a question about Opera the other day, just in passing, but I'm still a Firefox user, but I keep track of the other browsers and it looks like they're going to be, maybe including that soon. Yeah, I see someone says, I used Chat GPT recently at my job to write a script that mails me when a new file arrives on our FTP. While you're using FTP still, but nonetheless, awesome that it worked flawlessly. Just be really careful about if you're creating a solution in public view that the way it learned might have been from someone else's code or something and there could be issues with that. I don't really know how that's going to play out because you see these people are trying to sue AI companies and all that. But then again, just be careful that if you're using code that might have been written exactly the same way as someone else's program that could be a potential issue. But then again, I think a good counter argument is there's a finite number of ways to develop an if statement, right? Yes, exactly. Someone asked the question in the chat too, any recommendations for a 4U PC rack mount case? I do have one. If you look at my recent build for my XCPNG Ryzen, if you put that in, you'll find a build and you'll find a 4U case that we used. I liked it, it worked. I'm not particularly fond of any 4U case. I didn't research from a lot, my staff did and they chose that one. I can't give you all the parameters of what made them choose that case but everything fit in it, that was enough. I could give you one thing that could have been it. What I've actually encountered is that these will be basically ATX cases. So you get a motherboard from wherever it'll fit. So that part's pretty easy. The parts will fit in there. But where you'll run into an issue often is with the power supply. If you buy a case without a power supply, a server case without a power supply, it'll often be proprietary and you just won't find one that works. You could probably fit something in there but the fan might be in the wrong spot, they're very particular. The 4U cases generally tend to support, most often they do anyway, actual ATX power supplies. So you don't have to worry about that but absolutely look into that first before you buy it. And if you're trying to save money by getting one without a power supply, if it's less than for you, do not bother with that. Get it with the power supply, you'll save yourself weeks of waiting for a power supply from a very specific place to arrive in the mail. The other thing, when it comes to server chassis, I haven't looked at all of them. I looked at quite a few of them and I would say the best you could probably get at this point, if I'm being generous, is a score I would give a 7 out of 10 for probably the best of the cases I use. None of them are like, oh my gosh, this is the best case I've ever used with like the perfect cable management and perfect everything. I've never felt that way and I tried a bunch of these. They get the job done, they look good, you know, you can have blue ones, gray ones, like I've done in my videos, but on the inside though, it can be a little bit of a challenge because they haven't, you'll see what I mean. It's doable but if you had an average PC case in the past, that's pretty much the same thing here. Yeah, there we go. Someone said we did, somebody's write case GPT, your case choosing software. It's an AI to choose your case. That would be fine, they'll probably just tell you to build your own. Yeah, but I've seen someone else say they used to have GPT for IP tables recently. That's another one. You know, asking some of those questions is probably relevant if you're not sure how to structure some of those commands. It does formatting really well, so if you need your, I did have it produce some YAML files for like the network configs in Ubuntu and it told, one of the things I thought was interesting is when I had it create the YAML file for that, at the end after it created the code, it gets a little copy button so you can just copy and paste it, it then lets you know, once you've done this, you need to run the network apply. I like that it adds the extra context, like here's the YAML code, but by the way, those are those little helpful hints because if you change it, it doesn't change your network settings, it's not until you do the network apply settings, so it gives you some of the extra context on there for that that can be very helpful, like editing the YAML file is easy, creating one from scratch, you may get the formatting wrong if you've never created one before. Yeah, I just tried to get on chat GPT as we're recording it's at capacity right now, because what I wanted to do was just ask it to tell me a UDP joke and if it comes back at the response, it's automatically wrong no matter what it is. If it doesn't like close the browser tab immediately when I ask that question, it got it wrong. So, but I was just going to find out, but maybe I'll try that later. I was curious what it would do. It's fun to play with. Someone asked a question about Drive Pass True. I don't plan on making an updated video, so whatever video and documentation I have on Drive Pass True, and it's also documentation in the XCPNG forums on this, or their documentation has it. I don't plan on doing an updated video, but no, I would never use that in production either. It's fun for testing, but there's plenty of discussion in my forums about all the problems people have with it. So, take that with a grain of salt. Yeah, doing Drive Pass True is just, it's not standard. It's fun, hacky stuff you can do in the Home Lab. It's fun for learning, but you're also going to learn when it breaks. You get a different kind of learning. So, don't put anything critical on it. I did it as some experiments to show how you can do some testing, but I think I iterated right very clearly in the very beginning going, do not use this in a production environment. I'm just not a fan of Pass True in any way, shape, or form. I always feel like there's always a better solution that people have at their disposal that may not be immediately apparent, but I just don't, I feel like it's just too finicky, I guess is the best way to put it. It can be. You know, Wendell has a video, it's a good one called The Forbidden Router, and I do like the video, and he discusses the pros and cons of putting all of your eggs in one basket and building one machine to do it all. It's a fun experiment, but it's also a house of cards because if one thing goes wrong. Now, you can do controller Pass True and things like that, but you're still doing, you're taking and modifying it from normal, like, hey, this is how we're going to use it, and that's how you're going to see it in the business world to what home users do. I'm not going to say there's zero usage in the business world for it, but there's very little. It's very niche to do any type of Pass True. Generally, they create a bunch of the same servers in the business world, and we don't pass through things specifically. Now, with graphics cards and kind of being passed through, but it's usually done through SRIOV where it's built in, like it's a native, and you're using like the specific cards that are for it that are expensive. Kind of related, Serve the Home just posted an article about the hardware. I know a lot of you, if you follow HomeLab, you follow Serve the Home. They broke down the hardware that's being used by ChatGPT, the type of cards they use, the NVIDIA processing cards, and I like the way the article starts out. Many of you may not have seen these because these cards are hard to find in $10,000 apiece, and there's this many of them in each server, so each one of these servers has X amount of cards in it, so, but they kind of, they do a dive into what that hardware looks like. I wonder what the scalping price is on a $10,000 MSRP video card. Oh my gosh. Probably quite a bit. That's expensive. I don't think that's probably outside the affordability of most HomeLab people speaking for myself at least. Yeah, and Jeff from Craft Computing has definitely covered this a few times. When these, he's went down that rabbit hole, he's got a whole series on getting cards to pass through and all the challenges on that. He's probably got some of the most comprehensive when it comes to graphics cards. He's tested it not just on one hypervisor, but a list of hypervisors, so check out Craft Computing Series if you want to go down the rabbit hole with Jeff of all the different ways you can pass through cards. Yep, absolutely. If someone does have a $10,000 GPU, MSRP, in their HomeLab, let us know what you're using it for. I'd be really curious. I think the question here is directed at me as someone says, why have two terabyte NVMe's when you're mostly using NFS Ice-Cuzzy? And so I have the option of having local storage if I want it. And I have two of them because they're mirrored. So in my lab setup or that we're doing or it was somewhat production, when you can lab things out on a really fast local drive, it's local NVMe's are going to beat out even a 25 gig connection to your storage server unless your storage server's NVMe. So because our storage servers are not, unfortunately, all NVMe, you know, no one's bought me one yet. No one, no one sent me a super fast server. So if any sponsors are listening, wants to mail me an NVMe server, I'm here. But yeah, that's great. It's kind of one of the reasons why I actually bought an eight terabyte NVMe for video rendering for the cache and don't ask me how much it cost. But to your point, yeah, local storage is great. There's a reason why I bought it, but oh my gosh, an eight terabyte SSD is not for the faint of heart. At some point it's going to be like, yeah, that's $20. It's eight terabytes, it's nothing. But right now, oh my God, pretty expensive. I can really add up, but man, is it fast. And I think this is our last question here we'll answer, which is the person we kind of answered earlier, maybe they hadn't joined, but I thank you for the donation here. When is it better? One, to have more VLANs with dedicated purposes, i.e. NAS, DNS and NAS, or two fewer, but complete VLANs as interface for each server on each VLAN where access is needed with, no need to hit the router. So we kind of cover this up. You have the same question, I think. Yeah, it's the same person asking the same question. I think they probably didn't join when we answered their question earlier, but we won't go into the whole VLAN discussion rant we just did about what is or isn't IoT, but I would go with only as many as you need and don't overcomplicate your life by creating way too many. Micro segmentation is a headache. It sounds like a good idea, but once you start putting everything on there and making pinhole rules between all of these different VLANs, it's a management nightmare to keep things working, or it's the thing you wanted to do because you wanted to learn how to pinhole every little thing. There really is a lot of devices, and I've talked about this with Synologies and with TrueNAS, you can firewall the devices themselves and lock down ports, so you can help minimize the lateral, potential lateral movement just by putting the firewall on the NAS and switching it to on and locking down only what's needed and following principles of leave privilege because if you're going to be punching holes between firewalls, you may as well have some of those devices on the same because it may be more convenient. Just don't go crazy with micro segmentation unless your goal is to unless your goal is to learn exactly how every servicing you wanted to route through the firewall, which by the way, as you route things through the firewall, you're putting more pressure on the firewall for performance, so that's something to consider, especially when you say, and this happens a lot, I tell people don't route your storage because once you start routing storage through the firewall, you end up with a lot of problems. For storage, generally speaking, like your iSCSI, your NFS, and your SMB works best on the same network, not through the firewall. That's where you get a big headache. So is it practical to have something in a VLAN? Or is it cool? It's just cool. I mean, if it's the only VLAN you want to create, fine, have fun, but focus on the practical. It's going to be security. Maybe it'll simplify something. It has to have a purpose, you know, don't don't just check the box. Just I want to solve this problem. Would VLAN be a way to solve that problem? If the answer is yes, go for it. Yep. And someone once going back to my MVMEs and my Ryzen build, XTP and G support software RAID. So you set them up in a software mirror and they're mirrored. So they're good. Yep. Keeps easy. You don't need a special recurrent. Mirror, mirror on the wall. Who's the fairest RAID of them all? Oh, yes. All right. Feedback at the HomeLab show we love hearing from you. We do have some guest appearances. So we'll at least throw that out there. We are going to be talking about Linux disc shows in the future and doing some interviews and all kinds of fun stuff. So we got plenty more planned, but we like hearing from you and thank you for joining us. This was awesome. Yep, sure was. All right. Take care, everyone.