วีดีโอ

They will rarely fit without empty space. The drawers at my desk have a bit less than 1/2" of space on either side. You can print spacers to keep the plates from sliding in the drawer, and I keep tall narrow things in some of those spaces. I keep my calipers in there at my desk. Gridfinity isn't really about using 100% of your available space. It is great at organizing your stuff, and having a standard grid size means you get to benefit from hundreds of other people's custom bins.

If you swapped out my NVMe and SSD caches for devices that are 10x faster without telling me, I wouldn't notice the difference. The videos I edit are only about 15 megabytes per second, so even with 3 or 4 videos on the same timeline, I rarely need more than a few hundred megabytes per second of throughput there. I also haven't seen a game use more than about 350 megabytes per second while loading, and it doesn't maintain that for more than two seconds at a time. As for the specifics of RAID 0, I would prefer not to do it unless I had a really good reason. I run my lvmcache in write-back mode. That means there is data in my cache that may not make it to the slow backing disk for hours. I would want to switch to write-through mode to make sure all data in a RAID 0 cache would be flushed immediately, but that might slow down my writes considerably. It is exceedingly difficult to max out even a bottom-tier NVME for more than a few seconds at a time. Just about the only operation that I perform that would max out my NVMe drives is a large file copy from one NVMe to another, and I can't say that this happens often. There is almost always a bottleneck somewhere else that isn't the lightning-fast storage!

The best part is that your SSD cache is going to be as fast as your SSD. At least, my lvmcache cache is. If you stick a 7 GB/s 1,000,000 IOPS NVMe in front of a 5400 RPM hard disk, then the cache is going to be super fast. The only slow down you usually run into is if you haven't played your old favorite game in 6 months, and you have to wait for the cache to be primed at HDD speed. I thought for sure that a year after recording this video that SSD caching for gaming would no longer be interesting, because flash storage prices were dropping so fast. Now that prices are, much to my disappointment, going in the WRONG direction, some of us may need NVMe caches for a few more years than we hoped!

@@patregan Yeah, pricing didn't get better in a while now. I mean, at some point NVMe SSDs started to be a lot cheaper than what they used to be, but now games are a lot heavier too, so it's pretty easy to run out of space.

Brian and I are both bummed out about this! We haven't seen our friend 3Dwebe, the designer of this fantastic case, anywhere in quite a long time. We miss seeing him around, and we wish people could continue printing his beautiful NAS case! Brian and I don't own the copyright, and we don't have a license to sell or give away any STL files that we have. I truly wish one or both of us did, because I would love to see people continue to print, customize, and enjoy these cases!

I don't know for certain, @blufade! I am using one 4 TB hard drive in my homelab setup, or at least a 1 TB partition on a 4 TB hard drive in my homelab server, but it is mirrored to a 1 TB partition on another 14 TB USB3 hard drive. I don't think we talked about that in this video, though, because I don't think I downsized that from a RAID 10 to that tiny RAID 1 until almost a year after we recorded this video, so I am guessing that isn't what you're wondering about. I am guessing that you're concerned about the single 14 TB hard drive that I have hanging off that Pi at Brian's house. I am not concerned about it. There is a full copy of all that data synced to my workstation. There is a full copy of all that data synced to a USB hard drive attached to my NAS VM, and that btrfs filesystem has hourly, daily, weekly, and monthly snapshots. The absolute most important data is also synced to my laptop. When that 14 TB USB drive at Brian's house eventually fails, I can order a fresh drive, sync all the data back up to it, and head over to install the replacement.

@@patregan :D good to know. The 14tb hanging from of a usb3 cable connected to an rpi bugged me.

@@blufade Between the combination of how cheap Raspberry Pis used to be (and almost are again!), how inexpensive N5095 minipcs have gotten, and how HUGE hard disks have gotten, there is a pretty good chance that I will never use a RAID for my personal stuff ever again. Now that my data fits on one disk and these miniature computers are so cheap, I can make sure my data is spread around as far as I can get it. We have been talking about recording an episode about this! :)

wait, why? DIY PnP is freaking amazing.

There is a nugget of truth in what you're saying. The more states each flash cell has, the more fragile it is, so QLC is way less durable than SLC. Even so, pretty much any SSD sold today is more durable than drives were were buying four or five years ago. My first real SSD, an 80 GB Intel X25-M had a ridiculously short potential lifespan. The Crucial 480 GB I replaced it with was guaranteed for 72 terabytes of writes. The low-end Samsung 980 (no pro) that I replaced it with is guaranteed for 600 terabytes of writes (and S.M.A.R.T. thinks it'll manage double that!). This is my cache and my root/home drive, and I have only used up 4% of its lifetime over the last 16 months. The manufacturers know how reliable the flash is. They know how much flash they need to set aside in reserve to swap out blocks that are aging, and their wear-leveling algorithms get better every year. Even the garbage-tier $34-per-tb SSDs will likely last longer than the fairly premium Crucial M500's 72 TBW. I don't have nearly enough fingers to count how many mechanical hard drives I have had fail on me over the three decades. I have enough thumbs to count the number of SSDs I have had fail on me since 2009, and one of those was kind of my own fault!

Buy larger capacity RAM and allocate more ram to the cache. I hazard a guess that would take some of the load off of the SSD.

Everything is a balancing act. If my machines have more RAM, they will for sure use more of it as cache. I bet we could measure some improvements on my desktop computer here if we doubled or quadrupled the amount of RAM, but I don't think I would be able to tell you how much RAM you snuck in there on me without checking. Most of improvement would be too small for me to perceive. Not having enough RAM for your workload is extremely obvious. It is difficult to see the different between just barely enough RAM and four times as much as you need.

It is a super nifty concept for sure!

Welcome to 3D printing Ramona!

also, I am brand new to tailscale - probably obvious. sorry please help

It seems like TH-cam ate my reply, Snakes and Trails, I'm sorry! I installetd Jellyfin on my NAS and I was able to access Jellyfin from outside of my network using the Tailscale IP using Jellyfin's port number without any issues. I don't know why you're trying to set up port forwarding--but it's definitely not required to make Tailscale work. Port forwarding also wasn't necessary for me to get Jellyfin working. You can test if Tailscale is working by: pinging your NAS' Tailscale IP using the "tailscale ping", or pulling up the TrueNAS web interface using your NAS' Tailscale IP, or even just checking the Tailscale admin page and looking to see if your NAS is listed on your Tailnet. I assume that both the Jellyfin App and Tailscale App were installed from the TrueCharts catalog. If you need more guidance, you could try joining their Discord server and submitting a support ticket.

Thanks for getting back to me. I got it up and running finally with help from a raid owl video. Is everything normally slower with tailscale running when you are accessing your network remotely? Or could modem/router/isp be a bottleneck?

​@@snakesandtrails9290 The answer to both of your questions is "Yes, maybe." Tailscale forms an encrypted tunnel between endpoints on your Tailnet. There's overhead involved in encrypting/decrypting the traffic in both directions. And there is always the chance that your speed is constrained by any of the network gear between two devices.

Thanks Brian!

Yeah! The days of having to worry about write endurance went away a long time ago. Even the cheapest solid-state drives come with 200 TBW warranties, and there really aren't that many uses that will crunch through that many writes in a decade. I have been running the NVMe cache in my desktop for 9 months now. SMART says I am up to 16.4 TB of writes out of the 320 that would be under warranty. SMART also says I have used 2% of the life of the drive. That means Samsung expects the drive to survive two or three times the volume of writes as they warranty. The drive may fail for some other reason, but it would take so long to hit the warranty that all my computers with m.2 slots will already be long since retired! :)

@@patregan Wow that is REALLY good news. You can just swap out the SSD. Given that a 500GB SSD is so cheap, I really think you have a winning solution. The question for me is would I PrimoCache with a SATA 500GB SATA SSD or a 250GB NVME SSD - both of which I have spare. I guess that I could tier both of them, but part of me feels that I should choose one over the other. You do very similar things to me (videos, games, files) - so how much L2 cache would you recommend? Thank you. I loved the video.

@@SC-hk6ui I am sorry S C. This is my third try replying! TH-cam ate my long reply, and it ate my short reply with just the link, and they didn't even show up as a held comment! Here's another try at posting the link as the channel owner. Do you think they'll let me do this?! blog.patshead.com/2022/09/six-months-of-lvmcache-on-my-desktop.html

@@SC-hk6ui I left you a long reply S C, but TH-cam ate it. Hopefully I remember what I wanted to tell you! :) I think the most important thing I can tell you is that you can't do anything wrong. I wound up splitting my storage into two chunks each with their own cache. My NVMe cache is only 300 GB, and I realized this is too small for the amount of data I regularly deal with. I usually edit around 100 GB of video each month, and I have games like Borderlands 3 trying to eat up 120 GB of the cache all by themselves. I was getting into situations where video was pushing Steam games out of my cache. If I were to start over, I would use 900 GB for the cache, 100 GB for the operating system. I don't get to start over without doing a lot of work, so the lazy thing to do was put my old 480 GB SSD up as a cache in front of the video storage. How much SSD cache you "need" has less to do with how much total data you have and more to do with how much data you access. Even if I didn't split my cache, it wouldn't be a big deal. The worst thing that was happening to me was that editing 100 GB of video might push a game out of my cache, and I have to "endure" slower loading speeds one more time.

@@patregan Excellent reply. Thank you. I've subscribed and I'm working my way through all of you excellent videos.

I found it. TrueCharts is a separate app category list.

@@Saturn2888 Nice work! How are you liking Tailscale so far?

@@briancmoses it's great! Super easy to setup, but the transfer speed seems up and down. It was at 1000Mb/s in my local Internet to itself, but then it was later only 260Mb/s consistently since that original test. So I'm not sure how I feel about it.

I have only ever used taildrop with the command-line client on Linux. The documentation leads me to believe it is different with the GUI clients for other operating systems, but with the CLI you have to specify the target directory when you run the `get` command.

If I can't SSH to a server, I can't manage it or rsync files to it.

@@patregan Still something I might be failing to understand here. The tailscale service runs on my linux server (I also set subnet routing). Tailscale is also running on my smartphone; If I open a ssh app on it, I can connect to my linux server shell even using its LAN IP. Done! Why should I need this new Tailscale feature then? Thanks

@@jeytis72 Now I understand what information you're looking for! You can't just connect to your local LAN machines via SSH, unless you're only using basic password authentication. You need to create at least one private key, then you need to install that public key on every machine you want to SSH into. Tailscale SSH sort of handles all of that for you. Tailscale is already generating public and private keys for use with Wireguard, and it does this automatically for every node you add to your Tailnet. Tailscale SSH lets you use that infrastructure for authentication and authorization. There are other ways to have centralized SSH key management, but they're much more complicated to set up. If you're already using Tailscale as your overlay mesh VPN, then you get Tailscale SSH nearly for free, which has been awesome!

@@patregan It's clearer now. Thanks

I don't know if there's a simple answer. When you're giving advice to someone, documentation is sometimes extremely important. Turning on a Wireguard or OpenVPN server in TrueNAS is likely easy enough, but then you get to the part where you have to forward a port to your VPN server, and everyone has a different router. Getting a bunch of machines connected with Tailscale (usually!) only requires enough knowledge to install a program and log into your account on each machine, and everything usually just works. If the goal ends are connecting a handful of machines back to your NAS, Wireguard or OpenVPN are fine. Tailscale and other similar mesh VPN setups really shine when your needs start getting more complicated. I host a little storage server via Tailscale over at Brian's house. Neither of us had to forward ports. When I needed to do some work on that server, I brought it home and booted it up here. None of the three people that connect to that server every day noticed that it was in a new location. It just continued to work, and they didn't notice when I took it back to Brian's house. For me, Tailscale's node sharing is the killer feature. I don't have to manage VPN keys or passwords for friends and business partners. We just send them an invite to connect to one of our machines, they sign up for and install Tailscale, and they can ping one of my devices. It just works so well!

@@patregan I get it. Thanks for the answer.

Oh that is exciting! I don't remember seeing that he had a domain name with a website up when we recorded this! That's awesome! Thank you so much Nomads Galaxy. The description is updated!

@@ButterWhat No problem! The website was made long after that video was posted, as Kralyn moved the project to it's own entity, like Voron :)

That's awesome! I have an older discontinued of Harbor Freight tool drawers. If I'd have known a few years later that they'd be charging more than that just for the small box that sits on top, I would have bought two or three more sets! They're 27" wide and about 5' tall, and I got the set for $160 with a coupon. You have the right idea, Shari. I totally need to start printing Gridfinity organizers for that tool box!

VTX antennas are difficult to solder on correctly. If they come lose, I just assume the VTX overheated and fried itself. You're usually OK on a $50 VTX like a TBS Unify when the antenna comes off, but cheaper VTX modules will burn out when no antenna is connected. There are so many hazards out there. Breaking and losing quads is something we all have to deal with. I hope you get up and running soon Adaptimus Prime!

@@patregan Thanks Pat! It seems this will be one of those things I'm always learning from :)

Which router do you have?

@@sirdewd2197 It's a Turris Omnia

The thumbnail is most excellent!

Both systems are fantastic, Robert!

I am more than a little envious of all the Vorons my friends are building! :)

You and Thomas have me tempted, Jaryd! I am going to put this on my list of things I need to try in the near future!

Yeah he did! What are you going to do about it?! 🤣

I agree Devin! Brian is doing a good job!

Thanks a bunch, Devin! Are you thinking about setting up some ESPresense Base Stations?