 And welcome to a Streaming Saturday. I almost said Sunday. I think about that a second. I'm like looking at the title across there Yes, it is Saturday and I've been playing in my lab a lot Actually, I don't know what's the difference between lab and production sometimes when it's like I call it my lab But a lot of my video stuff is technically my Business in production. So yes And sure I'll run beta on my video production stuff because why not? And we're gonna we're gonna play around on some stuff. I figure I'm gonna play with it Anyways, why not do it live and people can watch what I'm doing. So I'll be showing that in a moment Not sure if you're here for the video, but I have a question. There's a two and a half Is there a two and a half inch SSD brand you'd recommend for boot drives for two nests looking for small No, one-plus TV 256 gig max, you know, I've Not for production systems. Well my lab systems, which I know are somewhat production systems Not stuff. We not stuff. We sell to customers. I'll say I've always Used whatever I had I literally And we'll just share the screen here. So on that topic. I've had any problems with the Some of the different brands of like sand disc or What's the other one? I know I just happen to have some sand disc laying around So if we go here and we look at the discs that are in this system right now, I literally found a pair of So this filter for the boot pool, there we go Are they this one's a Samsung? So one's a Samsung is the other one Sam at a sand desk I paired them together. So I made a boot pool out of these two drives There's one drive is slightly bigger than the other. So there's some leftover space when it mirrored them But whatever, I just it's a boot pool. So it's really not relevant to me in terms of that But yeah, it it's not that big of a deal I Have not had problems with Samsung's have not had problems with the sand disc ones, especially these smaller drives like this So if you have them laying around awesome, I do they're pulled out of systems. I don't even know they probably have quite a few hours on them But they're in a mirror. So I actually we actually have we just haven't thrown them out So I have a handful of these floating around the office and so I put two of them in here. It's actually Go and See how many is it tell me how many hours are on there? Hold on. Oh, I don't like that. It drops me like this. Can I just type sudo in this work? Well, once we put the password really long Can I put the password in? I'll never do it this way. Can I control V and we'll let me paste All right, well SSH in and I'm curious now how many hours are on that particular drive So if we SSH in a slash well drive was that Gotta go Go back over to system Whoops actually I can go here storage and dis Filter for the boot pool. We'll pull that back up. Let's go. I'm just gonna pull, you know, that's a feature they can add See if I run the smart test results are not in here I Think it'll give me well, we can run a smart test, but I don't think it will give me What I'm looking for in here. I from the command line. I can just do a smart control dash a and Let's do a manual. I see a short test Start let's see if this fills in the data. I'm looking for I Just want the answer. How many hours are on this drive? That's like a handy thing to know. Does it have that? 874 hours, so there's not a lot of hours on this particular drive. So Yeah, I mean if any reliable drive, it doesn't take a lot There's not a lot of read write going on on the boot pools So and as long as you're backing up the config file, they're easily replaced just back up the config file You know, I know people have talked about this when I did my config video They were asking like well, hey, you know, how can I automate this if you look? There's ways you can set up cron jobs to automate the backing up the config My general thing with my churnass servers is I only back up the config when I make changes because that's when it's important So I don't have any auto set up which seems to surprise people But maybe I think differently because I'm like, okay. I'm changing shares. I'm making modifications. All right modifications are complete They're validated. They work export backup save. I do this with PF sense I do this with really anything I configure that has a backup file once I'm done configuring it I've validated that the configuration is what I want I then back up that config file and occasionally I back it up prior to Just in case and so I have it at the at the ready and then I take that file and I put it to a structured Directory I have that has each device and then has a backup file in there And I actually have revisions going back years because this because they're all small so they don't take much So I don't really bother with the automation on it. It's more of a mental thing that I have because if you said the automation are you gonna validate that the automation is always working because now you start depending on a Haki at an automation because it's not officially part of true nas to be able to do this and Does doing that actually? you know Do you follow through do you make sure your automation is working and Validate that the backups are being done and test restore etc. When I know if I go into true nas and I just go and download something out of there. So you go to the system General and you can just go download file Save there we go now we've downloaded it. I've got all the configuration backed up. So, oh See Any experience with stuff on xcpng I do not at the moment it supports it though stuff is a supported option in xcpng There's some discussion in the xcpng forums on this But I haven't done any of it firsthand We're looking at your NASA run as a database torture For xcpng VMs is a way to get close to ha with true nas I know he's only available x hardware minimum downtime is okay How fast do you want to synchronize the data is what it comes down to the there's two ways to do this in true Nas one of them is through zfs replication. So zfs replication would allow you to Replicate the data. I think the fastest you can set it is one minute So you can set one minute. Yeah, you can set one minute replications There's option a option B And I don't know that this is great for databases Because of the way they work it might work. I don't know Maybe not is syncing is another option That's a way to do it. You could have syncing pointed at two different places and have them synchronizing together I don't know if that would do the database transactions properly Database transactions are kind of specific and it's best if you use tools built into the database for example, some database can have Where there's a master database and then there's subsequent replicate replicated versions of it that You know, you can do this with like my sequel You can do this with I believe you can do some of the Microsoft ones I think they have a special thing they use for it But you can just have the database in more than one place So that's that's probably a more logical option to do that I know as you post that your Nance had better NFS performance how this was achieved I I didn't read through every detail, but it's a updated kernel and they've been doing a lot of tuning and You know lots of optimization of code. I didn't look to see exactly how they made the code better I have my faith in them that they have been making the code better. So I don't really know I know it's in their document. So not documentation. It's on GitHub like they've updated what version of ZFS There's a lot of little things that well actually know if you're on 2310, you're using the same version of ZFS as 24 So it's actually probably not specifically ZFS, but it's open source. You can go through all the code changes that they made Couple cheap Kingston 120s Yeah, though those probably work fine too. Yep. True Nance is dumping Gloucester. That's correct There is enough that you probably don't want to use USB for the boot pool. That is correct I I had a USB boot one running all the way until recently and it didn't die I just finally said I should probably replace that it the USB drive was running for four years still working one question To the chat mostly has any of you tried Nutanix, you know, I will leave my Commentation comment here. I know people ask about Nutanix and feel free of course people in chat to chime in on this One of my problems I have with Nutanix is you're trading one proprietary company for another proprietary company You're not you didn't move the bar by a matter of fact specifically publicly traded private company that could also be easily bought by Broadcom because Broadcom's got enough cash to acquire Nutanix Whether or not they will today depends on if they think too much of their customer base goes to Nutanix. They might so Open source all the way so That's one of the reasons I'm not a Nutanix fan When running to your Nance at home primarily Nance a few small containers you recommend four or five or six drives thinking Z2 This is all about your risk tolerance You know, let's actually walk through real quick in a moment here I'm going to show you how I'm setting up one of my systems at home Well, I can show you how one of my other systems is set up that runs all my videos I have and for reasons it'll maybe be clear or not clear at all depending on who you're asking um I don't mind running because I have extra copies of all my data here at my studio and also two more copies off-site So I uh, I like to live dangerously and if you notice Let me zoom in here This is uh raid z1 And uh, I'm fine with that and here's more. Here's an eight wide raid z1 I'm fine with raid z1 because my risk Tolerance I have all the data going here And I have all the data replicating to another system here and then from there I have all the data replicating again to an off-site system That also replicates two more times. So there's two off-site copies There's two on-site copies of the data So I don't mind z1 So it comes down to what your risk tolerance is HA pro line microserver proxmox or xcp and g i'm on team xcp xcp and g for sure Hello, dakota. Hopefully life is going good over there on your side of the world Uh, what is the model hba card in etl 15 from 45 drives? If you don't you know if they're using more modern hba cards, uh, which have less bottle next try I don't remember what it is. Uh, I it's been mentioned in all the other people who reviewed it I believe said that in their videos. It's the hba card is If i'm not mistaken It's built it's built into the board. It's whatever supermiker has built in but I I don't remember the name of the chip that's on there. Um, maybe later I'll ssh into it when I finish my current projects and look at it But uh, you can also I think the 45 drives has it in the spec on their site Probably just google that so the homelab 15 Does it say? I know it's a super micro board fully assembled Does it say what yeah, we're here. So here's your board so super micro highlight copy search So whatever it's a super micro x11 sph n-ctf Does it say what controller that is? Oh, I hate this. It's a broadcom You know because broadcom so it's got a broadcom controller Everyone loves broadcom. I know anyways Uh, planning upgrade from core to scale could I expect replication tasks and cloud backups or back plays migrahy? They should be fine Uh, if you're speccing a switch of fabric topology, would you recommend as a core stack switch for a massive environment? 5000 activites needs to be stacking redundant PDU. Um, we have a 100 gig switch With a similar demand on it. Um, we're doing a rista. So I think a wrist is not bad for that You said at right time once tom gets the new tanix market Once do tanix gets the market share they will sell the broadcom. So You know, this is one of those things someone got mad at me for tried to tell me I was wrong and I'm like This is an article from a while ago new tanix for sale they have been trying to sell the company and uh The reason why and obviously I don't know what their status is today But one of the things you got to remember The people who invested in new tanix have been investing money into a company that loses money Because new tanix isn't profitable until now new tanix makes it first profit ever and that was in 2024 Probably related to you know speculation from uh with the broadcom buying. So I mean You got a big publicly traded company that has a bunch of debt that they owe to investors probably They're trying to sell and have been trying to sell and they just now made a profit So we're a world the tech world. It can be a little strange. I'll give it that but um Yeah, I don't I don't see a play that makes me like new tanix. I'll throw that out there Oh, you're right. The second one is raid z2 Yeah, the first one's raid z1. The second one is raid z2. You're right given the specs of the uh Given the specs hl 15 and half the back lane being seda I bet Half is run from on board seda from the chipset plus hba for the rest. No, it's all plugged into the hba Uh somewhere. Do I have a photo of the hl 15? Maybe I don't know other people do have photos of it, but it's all There is a photo I have where I took it apart I haven't done the review because I've been busy doing a few other projects, but I did take the hl 15 all apart I just got to find that photo Oh, there we go. So there's the hl 15 uh partially taken apart So you can see the connector down here That's the backplane plug-in so For those that seem to be very curious about that How does that orchestra do integrity checking for backups? Uh, not the restore method for proxmox backup server after a backup the data is scanned and hashed I was burned by zen when a restore failed I mean it validates It does a check someone it does the backup. So it does check some of the backups to validate that the data Matches that it's on there. Obviously doing your restore is the way to Absolutely validate things, but it does Validate the backup as it's being done First homelab recommendation for true nas system and support ice cozy nas functions looking to host VMs on another system And for that nas to be expandable later on as my system grows true nas mini r I really like the true nas mini r, but one thing I will say about it It's not that fast It's the one downside it is not a high it'll do 10 gig SMB got that covered No problem there, but if you want it to perform for nfs and ice cozy, it's just not that fast for those things That's the one downside of the true nas mini r Uh, it's not high performance. It doesn't have a super fast chip in it You know, I thought about because they're in people have complained and I get it if you look at the chip that's in As a matter of fact, let's go back over to here. This is my Uh, well, I have more than one of them. So let's which one's this We'll we'll log into this one here Figure out which one I'm in There we go These have that intel atom This is a this is another uh The true nas mini 3.0 and the true nas mini r both share the same processor. It's an intel atom And they're just not super fast. So if we search like I don't know Passmark There's the pass mark on that processor. It's not it's fine for an ass, but it's not incredible Therefore you're not going to get amazing ice because your nfs performance out of it. It'll work It just won't work as fast as something that's a little bit more performant Is it good if I add a bunch of hds as a vm guests and set up a read for those drives In the os to have a large drive. What is the pros concept up? Like you want to pass the drives? directly into The guest I mean you can it the reason I don't do a lot of that is frequently customers have more than one Host and you want to be able to move your virtual machines between hosts if you tie the guest os to Something directly in that host you don't have that as an option because it becomes married to that host So it's kind of a preference. I have but yeah Xcp and gv is an arc show what sort of complication should I look out for when I want to add a gpu for video encoding originally inquired a In video for um, I did my pasture video but the The challenge you run into isn't necessarily the pass through The problem may be run into is in I'm not an expert on this jeff from craft computing has done more videos on this topic than I have He is my go-to person I will reach out to jeff at craft computing or even watch one of his videos There are weird rules around the way nvidia licenses or does something with drivers I don't know the details I know there's bugs that are more related to nvidia drivers than it is anything else Where when you try to do the encoding it has some problem because of nvidia Being weird about things They're selling an all flash unit with scale. Do you know how they tune it to make sure iops are good It's more about how you set up the v devs is probably the most important thing to getting the iops good Second thing is putting really fast Drives in there that have good, you know enterprise like sass drives with good q-def They use nv dim on some of them as well to get better performance with the slog So it's not there's not any special changes. They make to true nas itself It's all in the setup of the systems deduplication is a disaster waiting to happen if you're not careful Uh deduplication definitely can cause some major performance problems There's a bunch of new code coming down the pipeline to make deduplication better But deduplication can definitely cause you a lot of headaches. I've seen people with We have people reach out to us. They want us to fix their deduplication problem. I'm like, there's no fixing the problem It doesn't scale very well Uh with performance the the more data you have the more data it has to compare against for deduplication And by the way, deduplication is only effective if you have data that can be deduplicated And not all data can be deduplicated. That is where the problem is So if you don't have data that can be deduplicated that there's no point even turning it on All right, let's go over here. I'm closing all these tabs What is I need to make a new storage pool here. What is this storage pool? Okay, I got some test data because I was testing these drives, but I got to rebuild it So we're going to go ahead and do that next Oh, you want to put a bunch of vhd together and Set up a raid the problem is the way raid works is It's going to start talking to each of the drives and then it's going to put pressure on the System to manage all those small rates between the drives So as long as you have enough compute to be able to do it It should work, but you just got to remember your The cpu. This is actually always the hang the big hang up processors. You can pass through relatively easy so to speak hypervisors don't Really kill processor speed that much, but virtualizing a hard drive definitely has a Cost so to speak of cpu cost and performance cost So sometimes when you do that you can watch the system Have the hypervisor itself start having a ton of cpu time Trying to sort out the raid rights and it actually ends up being slower I don't know if it's slower and that's going to depend on your system in your situation It's a fun experiment. It's not something I would ever run in production I haven't really had a use case for building raid arrays with a bunch of vdi's But you can I've done I I I should say I've done it for testing purposes like to validate something I wanted to test, but I've not done it as a Uh It's more of a thought experiment than an actual Something I would ever want to do in like production, but let's kill this and disconnect this pool We're going to destroy the pool We called this one test so we'll go ahead and Destroy the data on this pool We're doing it live. I'm glad I called it test or I would have had to think a little harder Just you know running through things All right, I need to build a new pool I know I'm going to call this one dozers. That's the name of this particular system. We're not doing encryption Now this is a feature that you won't necessarily see in true nast unless you have one of their hardware You can actually limit the pool to the enclosure If you have multiple Enclosures attached you can choose which one. I thought this is kind of neat. It really doesn't matter There's only a single enclosure on this one because this is a true nast mini But I thought it's cool that they have this enclosure option data layout We're just going to go z1. We just had that discussion, but I think z1 z1 should work Should give me enough we're going to go a width of five Who gives me a raw capacity of 50 terabytes? I guess I could go read z2 Gives me 38 terabytes a little bit of extra Extra head room in there Why not? It's only data Or do we want that? I don't know. I probably want that 10 extra terabytes. This is just going to be a backup server mostly Or do we I don't know see this is you're watching me kind of second-guess myself, right? Like I have enough storage I only have like 15 terabytes of data. I think I need to move over to this So it's one of those things like yeah It's one of those things uh, you kind of consider like how long will it be down for? Does it matter? It's a backup server. We're getting the z1 So we'll get a z1. I get this 50 terabytes. Then we get next I'm not worried about a log drive because it's a backup and uh I'm not worried about metadata cache. These are things that people worry about a lot. Um, that I'm good. I have some upcoming videos This is what I've been spending my time doing is validating a bunch of like how I can demonstrate how these log drives and cache drives Have minimal sometimes value Based on the circumstances and in this particular circumstance the workload that's going on this The log drive and metadata drive are not going to make any difference at all These contents will be erased. Hey, why not build the pool? yeah, um Wendell's video on this is great The intel cards are actually pretty impressive But I do I really I would recommend watching Wendell's video on that for sure It's uh, it's called It's the it's intel video pastures. He posted on level one text within the last week or so Maybe within two weeks ago It's it's a good video though. I really liked it Good tutorial on vpn with unify express, you know I'm kind of slowly diving back into some of the unify stuff because they've updated the firewalls so much I don't know anyone who's done one my You know myself, but uh, I will be doing some in the future It's always the challenge is always finding the time. I have a million things to do. It's finding the time to do them all but yay, we've got data and uh Let's set up some apps on this actually I'm gonna copy from one of my other ones Choose the pool Get these apps going Now this is not something you will necessarily have to do but I will show you a thing I do because the default IP ranges overlap with my office IP ranges and it'll cause weird I've I pointed this out to people that it will um Uh cause some problems if you have IPs that are in the range that it chooses under advanced settings You'll have a problem. So we're going to fix that real quick There we go. Whenever you do this it actually goes and rebuilds the uh Kubernetes on the back end So it it warns you that it'll reset all the apps. So you should do this When you first set this up for uh, setting up apps All's I did was move the IP to a range that doesn't overlap with my network Because you'll get it you'll get a system that uh, won't route properly because it has a kubernetes setup That has a range that overlaps with its main IP address Um, I've actually helped people sort that problem out a few times where they set up the applications and They ended up with that problem In you're actually watching in real time because nothing's sped up here the Slowness I guess you could say of that particular processor. It's not wicked fast So this I I mean I have no problems. I mean I'm doing all my video editing and everything else on this system here For you know All the videos I edit they're all 4k 60 is what I shoot them at so they're large video files. They work fine I can even bump it up to Pro res raw and edit right off of these it works it handles massive video files But certain things you'll notice that that processor that doesn't score very high Like when you're setting up the apps, it's not it's not as snappy as my my hl 15 is definitely faster Still initializing the app services doing the app discovery At the very minimum I want net data on here. Come on Sometimes you guys wait a second view all apps. Come on that data. There we go Takes a second for it to start indexing when it first sets up before it actually finds them install And after this gets installed it will start sending data to it Because the goal of this particular box is to get its get a bunch of data sent over here for An extra set of backups, but one of the other things that I'm going to be doing a video on soon is How replication works? Matter of fact, we're going to fix replication right now because it's broken Um, uh, maybe I won't fix that right now. I'll just do the replication to the other one But I do have an upcoming video where I break down how the replication works And then how to fix it when it breaks because I just moved a server So all this replication is not going to work properly. So if we go here to credentials And we choose our backup credentials I'm going to rename this one old Because that is now the old key Because I need to create a new key because it's got a new IP address Because you can change this here And now when I go over to the replication you look at which key it's using Oh, that's a snapshot There we go You see it says purple nas old. I need to point it at the purple nas new To make this work. So we go back over here to the credentials And we'll build new ones here to make that work systems credentials These are the things that People don't necessarily Whoops. I'm on the wrong. I'm on the wrong one. It's this one credentials backup credentials add purple nas Put the url in needs the password copy pasta copy password Put that in we'll say generate new Actually before I do that. I think I need to call this purple nas key old There we go Will it work or give me an error? perfect Now I can add the new one If you're wondering why it's called purple nas um That's because it's got a bunch of western digital purple drives in it Seem like the perfect name for it We want to generate a new key because I changed the keys on this one already Now it'll work Uh for users a beat a vmware euc products uh call error That seems bad Well, I'll I'll fix that later. I guess Huh. Oh, I know why oops All right. I got to fix some firewall rules. So is I I'm very implicit when things go across vpns I allow the things I want to allow. I do not blanket allow all the things so Okay, that's why that's got to be fixed for users of vmware euc products What do you think people should jump on now or wait till broadcom spun it out? Hmm I don't have an answer for that one What kind of vm resources do you all allocate for xo a service cp ram storage? Uh, not actually that much It kind of depends. I mean for the lab stuff I can get away with stuff that's really small for stuff that Is more production environment Um I mean I've got I do have 16 cores assigned to it so it can run more concurrent backups It's more cores that it needs. I've got six gigs four is fine But I went ahead and added six. I have a lot of memory in this system, but Yeah, you don't really need to uh go crazy with it You know, it kind of comes down to uh Test your test your workloads and see how well they work See you see where the processor problems are on it or if there are any Uh, any tips I wanted to look to get pf sense on how things are routed. I want to route wire guard to ip sec and tail scale I don't I've not tried that. Um It should work, but I know because some of them I don't know That's a better trunas forum question. I'm sorry pf sense forum question because I it's not something I don't have any times. I've had to set that up for businesses. They usually don't have me mixing the routing like that Looking forward to the replication video. I managed to torpedo a backup pool by deleting the extra snaps at the source Yeah, I'm going to cover what happens when you do that because we have a customer who's a frequent flyer The moment one of my technicians say hey, how did this happen? I'm like They they sent me a screenshot of I think these are the check boxes It didn't have the client's name and if I knew the client that did it I'm like, I know a client did that he goes. Yeah, they did it again Uh, I'm going to cover all the things that can go wrong and then the client watches my youtube videos Oh, that's how they found me. So I will hopefully I'll just send that to him like hey, you know, this is how This can break and this is how you can fix that Um, I have a video called like zff zfs data sets explained um Can what is the difference in ui terms? I mean the data set appears like a folder, but a Data set is a folder with extra features It has all the extra zfs features tied to it and everything nested under that data set inherits all those features folders are just folders They don't have any special features What is your favorite tool for ipmi monitoring and control commercially we use ovic so I don't know I don't really have a need for that. Um with with the open source infrastructure that I monitor But let's do something else real quick here I want to kick off the replication task Just that's actually the last thing I wanted to do like that I was going to show people because it's something that needs to be done and this may um Offer some insight. So let's go here to replication tasks. I don't need any new snapshots. I already have these tasks going right now Matter of fact, let's go to this one. Share this tab data sets We're going to go here add a data set. I'm just going to call it backups All right, we got the backup created here Go back over to here I don't need this test one get rid of that And because these already have tasks these are going to keep sending data to purple I also want data sent for example. I want all of my video archive We'll do that one first just because it's here So we want to uh do an advanced So ts vid archive to dozer And we'll walk through what it looks like to do. I'll zoom in a little bit make this easier to read We already have a connection called it's a dozer. That's already working Uh ssh plus netcat definitely the more ideal way to do this Um, I'll why use the video I'll explain that better, but ssh plus netcats essentially It uses ssh to establish the talk between the two servers Then it uses netcat to stream it instead of cap encapsulating it over ssh If you do this, it's a lot slower because you're you're limited to The encapsulation of ssh when ssh plus netcat You know, these are these servers are local to each other So this works great as long as they're local to each other or if they're reachable across a subnet That'll work Source Is going to be the uh lts video archive destination We want to put it in that data set backup and we're gonna call this lts video archive Simple enough We need to Attach it to a task. So this task is called video archive. Perfect. That's the only one we need to do Uh snapshot retention policy Make sure this is checked same as source is my preference You can do your own custom if you want, but same as source. That's uh an important one So whatever retention policy was building your snapshots also gets pushed over there I have seen people kind of forget about this when they do advanced replication and what you'll end up with is a Uh problem where it will just keep As they get deleted off the source, they're not getting this deleted off the destination and you can end up with quite the headache situation Uh set to either start replication task immediately after the linked periodic snapshot Complaints or continue to create a separate schedule replication I'm fine with just leaving it linked to this one here So this allows you to have one snapshot but two replication tasks that send your data in two places So we're gonna ahead and hit save Now one thing you'll note is depending on speed of the machine that isn't showing up If we click off of it, it will show up eventually. It'll it'll slowly refresh the screen or you can go backwards and Go back to the tasks and they'll be there and we'll kick this replication task off and we'll say run now Cross our fingers that this works See how fast it sends data over by going here Is it sending? Yes, it is open up net data Because we like pretty graphs Yay data What are we getting about? Five not bad Four four and a half four point seven That's not that's not unreasonable for spinning drives sending data to other spinning drives Um, what do you use to document they have set up in your home lab and I used to set things up a spreadsheet? It's simple I I just like putting things in spreadsheets I I keep saying I'm gonna set up net box. I just haven't Net box is nice, but yeah Been watching some old talks by uh, brian cantrell and development's dfs really Uh appreciation for the engineering that's on microsystems. Oh, yeah, there's um Man, I found this one video. It was from it was from one of the the um Conferences wow It was just how much engineering goes into how they you know do the snapshots how they Condense them back down It's like snapshots are so simple and easy to use because so much engineering went on the back end to make that magic happen Which I think is cool I'll give it a look. Uh youtube doesn't allow great formatting I had a channel data set had data, but the ui didn't show it and it seems dfs Uh mount itself was wrong. Huh? Can you do oh, I have no plans right now to do this I can't think of a use case where I would want to use A demanding environment of os a or a more complicated config not demanding really it's really just more complicated configs Where I would end up using uh os pf of the unify generally speaking I I don't know I don't have that on my roadmap or I have a lot of other things I'm gonna do before I get to that video. I should say I I Like that unify is trying to get some more professional stuff inside there but um Yeah, I I still if I have a more demanding environment I'm more likely to use a pf sense than a unify for something like that I've had child data sets not show in the scale ui If those data sets are created by my way, you can't create a data set by windows That's you're you're not understanding how that works The way data sets are data sets folders under data sets are not data sets so if I go And create a folder somewhere underneath a data set. That's not a data set data sets are different Uh, does xcp and g support usb nick. I have no idea and I wouldn't recommend it Uh, sd wan is a marketing term. So which type of Sd wan marketing term are we talking about sd wan means a lot of things That's why I actually did a video on sd wan Explaining all the different things that means in different companies in different scenarios So I guess it kind of depends which sd wan are we talking about We do use some muraki. Are we using some of the muraki sd wan features? I think so I probably have a few clients uh on it Uh, based on your statement this c 3 7 5 8 into lcp might lead to slow performance for ice guzzy on the system What cp would you recommend as a minimum requirement for achieving stable speeds? I would say you can achieve stable speeds with the three c 3 7 5 8 I don't know what stable speeds means Can you achieve Two three gigs sure maybe can you achieve 10 gigs over ice guzzy? Probably not Uh, so that's what it comes down to It it's actually one of those things it's really hard to calculate those things because it also Will it give good speeds if you're using a larger block and streaming the data versus Will it give good speeds if you're using a database application that's doing 4k chunks? Now you've gave it a different task. This is actually even when I ran those benchmarks and I tweeted them the other day Uh, let me pull this up real quick Where is There they are I was talking about the performance differences here. You gotta remember we got a 15 percent um performance boost on random write at 128k but if we go down here to the sequential write At one mag it was only 11 percent boost and when we did the sequential read we only got a 2 2.9 percent boost on the new version Every workload has a different profile for where it will or not Matter of fact, if you go look at one of my older videos where I was comparing core and scale It you'll pull your hair out because you're going wait a minute This application is faster on core, but this application will be faster on scale and I'm like yes It's core had an edge when you were using a very Specific workload, but that same workload won't perform as well on scale But what if you have a mixed workload now? Which one do you choose? This is one of the reasons that there's so much storage consulting we do because it's not an easy task You you really have to narrow down the workload and then you got to split the difference If they have multiple workloads going for the same system. So what's the best answer? Well, it depends and I know no one likes it depends, but it kind of depends on your workload And you may find because I have an iSCSI on this system the this C 3 7 5 8 or the trueness mini r that's the iSCSI for my windows steam library And it's actually quite fast So if your if your goal was to connect it to a windows system and use it for your steam library It turns out it's really fast. I get great performance out of it for that use case Uh, I've whittled my vmware alternatives down to hyper v new tanix oracle linux kvm and xcpng mainly to cover the mix of customers that we have Uh, some use hci some don't Hmm My complaint consistently Has been about open sensors are slow on security I I've listed this out if you go lauren stop video I can actually throw this whole thing, but it's lauren stop video slash open sense And i'll throw a link in here. This is in my forums. It's open. It's an open discussion Feel free to join in but I I point out all the times that they lag behind on security updates for open sense They because some people tell me that pf sense doesn't get as many updates as open sense So I investigated and what I found was they update for the sake of updating they have very frequent updates But they're not always for security patches because sometimes security patching is hard and they They rely heavily on Things that bsd happening. They don't really contribute back to bsd at open sense. They just pull downstream I don't think they really have as much engineering or don't appear to have the same engineering level team That pf sense has so i'm partial to pf sense, but I always say this in the end use what makes you happy If you decide open sense is what makes you happy I use it I like to think of a data set more as a mount point and not a folder um Except we actually have a lot of clients because they have different Rules that they may want to apply to different folders. We may have one base share For a company directory, but we'll have several data sets to the customer. It all looks like folders But there's different retention policies that are set on those data sets that are nested under the company share so yeah That's why I did that. I did a video explaining that so Bad wording true nests made saw a data set but showed 119 equal use looking at these at the path Showed my files same with the nfs export for the parents size screw not the chow Now this is actually an interesting one because It takes a while for true nests to show that one of the things I I might want to do Jump a bunch of data to true nests And show how long it takes true nests when you delete the data before it actually catches up with Knowing that it's been deleted. So, uh, that's actually one of those, uh Fun little things that you can run into in true nests Hey, look all the data is still For oh good. We're getting into five gigs now I am well and thank you for asking Uh, what is the correct way seven data set that is set up as an smb share and connected to a container? Have done this in the past anybody get the container turn on it's to turn off the share There's no good way is not the problem. I don't have a good answer um Because they've gotten better where you can use mixed permissions and they're fixing some of that in version 24 I haven't tested that much because I don't ever do that So I know they're fixing it. So I'll do some labs where I Do a video on it, uh, because they they've changed the way they create shares in version 24 now So log back into I think it's a system this is run 24. I have too many true nases open. All right. Yeah, this one's running 24 Uh, when you create a share Or create a data set that's going to be a share I should say So if this was a test share and we choose it as such like an smb It'll actually create the share automatically. So they've automated that They have a multi protocol to do this But i'm a little fuzzier when they do this with the apps because it's a different permission set I don't know exactly if it works properly or not So Yeah It's it's been one of the challenges of hey, how do I create a share and have it work properly? I probably should do some more videos on it. My my biggest hang up of doing videos on it People I get angry comments on sections of videos I did on my older true nas stuff going. Hey tom. This doesn't work anymore. You should take this video down I it's only a section of the video that's wrong because It well, it's right if you're using that version of true nas But you come all the way to the new version of true nas now. It's wrong Read through their documentation. Hopefully they keep it updated and have an official way to do it Oh, let's see. What else we got? yeah, open sense, uh, they do have a lot of updates which That only gets more confusing to me because they have updates But if you pay for open sense if you buy their business license version, they do have less updates But open sense was even more behind on security updates for their business version. So Um, yeah, I'm I'm very security first on things. So I'm that's one of the reasons we we I mean, we're using it commercially for business It's not just something protecting my lab. Uh, so yeah Um, I doubt I'm going to do any videos on vector packet processing I Lack of demand is one of those things like I try to do videos that help out the community I don't know How much that would help out the community. It's novel pf sense has some videos on it and talks about it Um, but I don't know that there's not many people asking for it Let them be angry. Jane you feelers. I I'd be lost today's really, uh Did you use open these know years and years and forever ago Maybe 10 plus years ago is the last time I really looked at, um open wrt. I know the project is still going Awesome if people are into it, uh, but I think it's actually it's one of the other firewall projects I'd say for home users is worth looking at commercially. We don't use it I don't plan on using it commercially, but I I do like stuff that helps out people in the home lab You know, it it's almost things like I I took all the business videos we did and there's always a link. You'll find Lawrence dot video slash biz I split them off the channel and I still we still do a lot of the business videos So we still talk about some of that stuff But I don't talk about all the things we do in business because I just don't think there's a lot of interest in it Uh, you know, especially like I I might do a video about the arista switch when we get all that set up for the client Because it's a hundred gig switch. It's cool. You know setting up all this hundred gig equipment, but the um I'm also someone asked me like I don't do a lot of muraki videos By the way, the our company not lauren systems, but the c and w r which is the it side of the company c and w r is a big muraki seller We we have a lot of unify out there They came over from the lord system side and they did unify already at c and w r be prior to our merger But they also do a lot of work with muraki But I don't I only I might do a couple muraki videos talking about it But I'm not going to do a bunch of muraki deep dives either I don't think that it'll help the home lab much and it's really just it would be more like an Advertisement for muraki and I don't love muraki at all. It works But I don't think they're amazing Um, I'll actually refer back to jeff over at craft computing He did the hundred gig video already and he has a breakdown of all the problems There's so many layers in between you'll run into that won't go As fast as you thought they would uh when you're moving data. So Uh, I I can't describe all of them off the top of my head, but good news is jeff from craft computing I think it's just if you type in like 100 gig. It's one of the few 100 gig videos jeff did I was going to use unraid for my dedicated home lab until I got the new pricing Did they raise the price on unraid? I don't use unraids. I don't really keep up with it Yeah, the gli net ones are base Uh, there's a lot of things that base on open wrt, but then kind of add their own sauce in there with it Huh? I mean, I don't I don't use unraid. So unraid Pricing It autocompleted two changes I don't know what it was. So is this more? um Basic plus pro So is that more than it was? I I don't know Oh, okay. So they went from a one time to an annual subscription. Got it That's the change. Oh, that's current, but they're moving it to subscription Okay Hmm Well, I mean This is the problem You know, it says right here pay once own forever Works in concept like Idealistically, it seems cool until you go. Well, how do we keep developing this product? And that is the challenge you have to have a business model that supports it now trunas has a business model by um Selling ix systems hardware. They fund the development of trunas. That's they have that's their business model It's not hidden you get trunas for free And they're hoping you buy some of their hardware to support the project A pf sense works exactly the same way net gate hopes you buy some hardware So you will continue the support of the developers that develop pf sense This is there's some fundamental problems in the market right now where uh, people get real upset when companies want to You know have some type of recurring revenue, but it turns out if you Don't want the project to come to a halt and go. Well, that's the last version ever then I mean, i'm not hating on them for uh trying to come up with it because I don't know any other business model unrade has except for Uh selling the license and if you're someone who's going man I've been using my unrade with my one time lifetime license for the last 10 years You know at some point developers can only get so much of the new market I don't think I don't think unrade is near as popular as uh trunas So I don't know I get why they're doing it, but I don't know I don't know a better model for some of these as part of the problem I mean, what are they how are they supposed to continue development? It it takes a lot of money Uh to develop products Especially because we are as we were discussing earlier in this live stream talking about security How do we want to secure things and stuff like that? Well, we'd love everything to be awesome and secure But uh that does take people Writing code testing code validating code to keep things awesome and secure Yeah, it's It's just low on my list. Um To test I have I have so many things i'm working on right now one of them that i'm working on is I have in my lab a Too many windows open Where did it go? Here it is uh a series of xcp and g machines that i'll be doing demos on so these are Uh, this is doing nested virtualization if you didn't know yes, this is supported You can do nested virtualization and xcp and g so I nested xcp and g and xcp and g so I can do demos of what happens when you lose servers Or more specifically when you lose hosts So if we go here And pull up the lab hosts or maybe these aren't working right now. There we go now they're working So if we go to hosts, we have these three hosts So i'm going to be doing a series of videos on How to fail over the host what happens when the host fails what happens when a server dies suddenly And you have to fix the servers. So that's uh I'm gonna it's basically an admin tips and tricks video to just cover a lot of those things Uh, what's up with the network video just finding time to do it That's really insert name of any video you asked me about is finding time to do it. That's always the challenge Um, I'm the only one producing the videos I'm the only one recording them and editing them the editing doesn't take me long It's all these setup and recording that takes all the time 80 percent of my video production is the setup demo script recording putting that all together Uh, because I do all the work on the front end. It makes the back end the editing only takes a few hours Uh, so it's just a matter of finding the time to do all those things Uh We'd like to see nest advert for windows hyper v running on xcp and g not likely Automate yeah, I just need an ai version of myself to create the videos. There we go. Um That we're getting closer. I've been playing with 11 labs. It's a company that does voice systems Uh, I've trained it on my voice. So can I just start cranking out content in my voice? Maybe there could be a future of that then I could just make all the videos like I could just write a script That I tell the ai prompted version of me to read the script and do the tutorial on everything So I just have to create the tutorials, but then I have to do the screen recordings So how do I get it to do the screen recording tutorial at the same time and explain what needs to be done? I think we're pretty far off from actually making that happen Yeah I mean I I don't I only do a limited I'd be a reason I sold my business so to speak and not really so merged if you will The reason I split my company apart to media production and then the it side Was so I could spend more time on media production. So I've been able to produce more videos Which is great, but It's not easy and I've I've tried a couple times to Bring people into the fold finding someone else who knows the stuff. I know and can do it on video That's the trick. It's not I've trained plenty of staff members who know what I know Who know how to set up true nas and know how to set up virtualization, but they're not good at teaching people how to do it so um Scaling this has been really really hard finding You know is a lot of times if people are really good at doing this stuff They just start their own youtube channel and do it. So I wouldn't mind There's at least one person i've been bugging because I think they'd be good at it to like join my channel But they they got their own channel and want it. They're making a go of themselves, you know, I've I reach out I ask I don't mind someone helping me out doing it. Um There's some things I could hire out Eventually I will like setting up net bird I mean I set it all up that took some time And I wouldn't mind hiring someone that had a task of building out some of the labs and making good notes for me But I kind of like doing it myself because it gives me a better understanding of it And that's where I'm kind of torn on it. So There's still some things people could do to give me the information and I can validate it So yeah, I don't know There is an in-tune lab that runs hyper v. I want to set up. I haven't had time to do it yet. Yeah, that's the thing Would you actually do it? Uh, or do you feel it would lose the It's sense of feel. I haven't seen any AI system that doesn't have that weird Uncanny valley to feeling to it. I don't know. I haven't seen any that made me go. Wow. This is the answer Yeah, teaching and being comfortable and on video Yes, it is because the technicians are that work for us are really really smart. They're great people It's just not In they I mean they teach their co-workers how to do things and stuff like that, but they're it's a lot different um than actually Putting a video together and putting everything that goes into Creating a tutorial. Uh, it's not simple is the best way I can describe it I'm not understanding your clinic. Hey guys, one thing. What about me? I'm not an AI I'm a real human being I don't understand the question. I think Uh But I think it's time to go watch tv now. So what we've got here. I went an hour and 10 minutes so the uh Fun of all this is definitely there. I definitely had a fun time playing with all this I didn't get as much xcp and g stuff as I was going to talk about Uh Yeah, I I will get this video I'm going to do a whole admin video because not everybody knows how all the stuff works in xcp and g Not there's like so many things have been added over time I'm going to do like an admin tutorial how to solve certain problems how to Uh cover when this or that happens in terms of like functions and things like that matter of fact, um Even things like this like what is smart reboot or you know forest reboot what happens with maintenance mode What happens if you detach a host or reattach a host? How does that work? I'll cover I'll have a whole video where I basically cover on cover all those little things like those little tips and tricks for admins I made a whole list too of command lines that you may want to use inside of a lab That's a whole another one. So, uh, that'll be I'm just it'll be a dedicated tips and tricks video. Essentially is what I'll be doing Uh, then the other video I am gonna uh See oh someone said Lawrence say hi to my daughter show it's live. Oh, hey, hello Um, the one other thing I'm going to be doing is some disaster testing with true nas I'm going to pull drives out of a running system I might do that as a live stream where I stage a bunch of stuff so I can show you in real time How true nas is resilient? Um against failures. So uh, that's on the list too, but hey as always love hearing from all of you Leave some comments Uh email, I've been throwing this up here You can email vlog thursday at loren systems.com if you want to reach out to me directly Uh, join the forums. Uh, loren systems.com is also where you find whatever socials I'm connected to so I'll go check those out there But uh, I'm gonna go chill out and watch some tv man. Thanks everyone for joining. Take care