Behind me, the door was open on my 3D-printed do-it-yourself NAS server, and it captured my attention. Above the power supply of the 3D-printed NAS case, the MK735, is a small cavity with a shelf on it. Over the years, that shelf has gone largely unused. Finding a way to stash some more storage in there seemed like an excellent opportunity.
I mused out loud, “I wonder how many 2.5” SATA SSDs I could squeeze on that shelf?” I grabbed my calipers, opened up OpenSCAD, and deduced that I could squeeze up to four more SATA SSDs in that empty space. With at least 6–7 unused SATA ports in my NAS, it sure seemed like I should be able to put this space to good use!
But why would I want to squeeze 4 more SATA SSDs into my NAS? My primary pool is at about 50% capacity, and in recent years, I have not been accumulating data at a rate which would suggest that I needed more capacity any time soon.
Instead of diving headfirst into obsessively 3D-designing a modified version of the MK735’s power supply shelf, I first started searching around for models which attempted to solve this same problem. These models would either serve as inspiration or maybe even get repurposed in my own design! After about half a dozen different searches, I found these SSD Stackers on Printables.com and immediately knew they were perfect! With enough 2.5” drive screws, you can create a custom-sized stack of 2.5” drives. These SSD stackers are fantastic!
I decided that I would use the SSD stackers to attach all four of the SATA SSDs to each other and then I’d put some screw holes in the shelf to screw up into the drive on the bottom of that stack. The biggest challenge in this was that the honeycomb design of the shelf is achieved by telling your slicer to put no layers on the top or bottom of the object. I needed those solid layers in order to put some recessed screw holes into the shelf itself. I wound up designing a second model that sat atop the shelf to give the screw holes some reinforcement and to raise the bottom drive up off the shelf a bit.
Altogether, it took two iterations to get this into a usable state. On the first iteration, I had some issues getting the shelf placed inside the case, and I felt like all of the drive cables were too far forward and contributing to the cable mess inside my DIY NAS.
My motherboard, a Supermicro X11SDV-4C-TLN2F-O, has a total of 8 SATA ports: 4 by the way of the onboard SATA ports and an additional 4 via its OCuLink port. Two of the SATA ports were empty, and I had never used the OCuLink port for anything, but had an OCuLink to 4x SATA cable from when I originally bought all the parts for my NAS. So quite confidently, I ordered the only hardware that I thought I’d need:
Then I ran into the first problem! I set the jumper on the motherboard to switch the OCuLink port to SATA and plugged all of the 1TB SATA SSDs into the motherboard’s OCuLink port via an OCuLink to 4x SATA cable. But for some reason, it wasn’t detecting any drives. Out of curiosity, I removed my 8-port HBA and even bought an Icy Dock M.2 NVMe SSD to U.2 2.5” SSD adapter, used it with an NVMe SSD, and a OCuLink to U.2 cable but with the same exact result. For some reason, my motherboard’s OCuLink port simply wouldn’t detect any kind of drive I plugged in to.
Undeterred, I immediately went out and purchased a new 16-port HBA. All too quickly, I had purchased an LSI 9300-16i to replace the 9220-8i that I had been using. Assuming that there wouldn’t be enough airflow in my MK735 to cool the LSI 9300-16i, I printed this neat LSI 9300-16i Noctua NF-A9x14 Fan Shroud on printables.com, ordered a Noctua NF-A9x14 fan, and assembled all the parts when they showed up. I also picked up four SFF-8643 to 4x SATA cables to populate the HBA.
In my haste, I overlooked a really important and painfully obvious detail: the MK735 case is not tall enough to accept full-height PCIe cards, and the LSI 9300-16i was not going to fit!
Ultimately, I wound up buying an LSI 9305-16i (I even briefly considered an LSI 9305-24i, too!). The low profile LSI 9305-16i fit fine in my MK735, and I was pleased to learn it was less likely to need the additional cooling that was strongly recommended for the 9300-16i.
I decided early on that I wanted to switch from a mirrored pair of SATA SSDs to a raidz1 pool of SATA SSDs. Unfortunately, that meant I would need to create a pool from scratch. Thankfully, I had a backup, enough hardware redundancy, and enough risk tolerance that I would be able to build the new pool and copy my data over from the old pool, but only if both the source and destination pools were in a degraded state.
I expected that this would be the most difficult part of my upgrade, but it wound up being easier than I expected! Shuffling my drives and data between pools seemed like the DIY NAS version of three-card monte. Once I had all the new disks plugged in and detected by my DIY NAS, I did the following:
fast
pool to my off-site NAS.fast
pool and waited for the resilver to convert it to a 3-way mirror.fast
pool.faster
out of the three unused 1TB SSDs and a sparse file.zfs send
and zfs receive
to copy my data from the fast
pool to the faster
pool.fast
pool and removed its related TrueNAS configuration.faster
pool with the unused 1TB SSD.faster
pool to fast
by exporting the faster
pool, imported that pool as fast
, and then exported the fast
pool.fast
pool from the TrueNAS UI.fast
pool was exported.
fast
pool.fast
pool.fast
pool to my off-site NAS.I was pretty impressed at how well all of this wound up working. Especially when I considered that I had never attempted many of these tasks before. Thankfully, everything went really smoothly. The riskiest parts (when I had no hardware redundancy in either the fast
or faster
pools) didn’t have hiccups. And thanks to the speedy SSDs and mostly-empty pools, the entire process of moving data around and resilvering pools finished in mere moments.
In the end, I accomplished everything that I set out to do. I was able to retire my unsupported TrueNAS boot pool partitions and I wound up growing my fast
pool by 2TB. But in that process, I managed to spend over $700! A good chunk of that $700 was spent on hardware that never was installed in my NAS or didn’t wind up staying in my NAS for very long.
I imagine that I can turn around and either return the hardware that I don’t need any longer (the 9300-16i, the Noctua fan, the M.2 to U.2 drive adapter, etc.) or sell it on eBay (my old 9220-8i and its cables) to lessen this upgrade’s financial impact. But even then, the cost of adding that 2TB of space wound up being wildly expensive.
What do you think? Have you ever spent around $350 (or more!) per terabyte to upgrade your own NAS? Tell me about it in the comments!
My loss might be your gain! Rather than return some of the stuff that I bought, I’ve decided to list them on my eBay store as well as some other parts that were made redundant by my upgrade!
There’s no disputing that I spent quite a bit of money and didn’t seem to get very much in return for what I spent. Regardless, I still enjoyed doing it. In addition to gaining those 2TB of usable storage, I also managed to add room for four more 2.5” drives in my DIY NAS, andcreate the possibility of an additional 4 more SATA drives via the HBA upgrade.
Part of my rationale in spending this money was anticipating what my NAS will look like in 2 or 3 more years. At some point in the future, I fully expect that I will be ditching hard drives in favor of SSDs. To accomplish that, I was going to need to be able to increase the total number of drives allowed by my NAS.
Once I can regularly buy quality 4TB SATA SSDs for $150–200, I will start slowly buying them. I’ll grow my SATA pool by swapping the 4TB SSDs in for my 1TB SSDs. By the time I’ve collected enough 4TB SSDs, I expect that RAIDZ expansion will have been released by OpenZFS and adopted by TrueNAS SCALE. This will allow me to grow my SSD pool to exceed my data storage needs and begin my move over to an all-flash DIY NAS!
Today, I might have a tinge of regret about spending $700 and only managing to add 2TB of usable storage. But down the road, I am confident that I’ll be glad that I spent it!
]]>Not too long after buying the hardware for my off-site backup NAS, I learned of the Topton N1 NAS. I immediately bought one and wrote a review of this toaster-shaped NAS over on Butter, What?!. I liked everything about the Toaster NAS. I liked it enough that I really wanted to try and sell it on my eBay store. Except the window on that particular product was closing, and I missed out on an opportunity to resell them.
Last time I placed an order to restock my eBay store’s inventory of motherboards, Topton asked me if I’d be interested in a product similar to the N1 NAS called the Topton 2-bay R1 Pro NAS.
From the information they sent me, it looked like exactly the same enclosure with an upgraded motherboard. Encouraged by my experience with the Topton N1 NAS, I bought a handful of the R1 Pro NAS and began to think of ideas for them.
From the beginning, the specifications of the R1 Pro NAS had me quite intrigued. I think the specifications were really ideal for what I’d consider an amazing “small” NAS or even an entry-level NAS for the budding data hoarder. The R1 Pro NAS, some quality RAM, an inexpensive NVMe SSD, a pair of hard disk drives, TrueNAS SCALE, and a little bit of assembly are all the necessary ingredients to build a very capable DIY NAS!
By far, my favorite detail from the R1 Pro NAS is the Intel N100 CPU. For its price and power consumption, the Intel N100 is a fairly powerful CPU. I am also excited that hard drive capacities have reached a point where a single drive is more than large enough to hold all of my data. These large hard drives have brought us to the point where we don’t need to build pools made out of multiple disks just to hold all of our data, and we have the option to focus on using these disks to add redundancy. Lastly, with a pair of 2.5Gbps network interfaces, the R1 Pro NAS helps clear the way for the DIY NAS builder’s most likely bottleneck: the network.
The icing on the cake for my media-hoarding readers is the Intel Quick Sync on the N100’s built-in GPU. There are quite a few people sharing their experiencing transcoding multiple streams on the N100’s GPU without troubles.
… then this isn’t the DIY NAS for you! The R1 Pro NAS’s outer shell and drive bays are all plastic. I personally don’t have any issues with its plastic, but I know that this much plastic can be a turn-off for some users. When I took the case apart to install an M.2 NVMe SSD and the RAM, I noticed that I was being very delicate with the case. Something about those plastic parts made me unconsciously worry about being too rough and risk breaking them.
I disassembled and reassembled the case several times. I put drives in and removed them from the drive cages several times. Nothing broke, and nothing even made me worry that it’d break. At its price point, I think this much plastic should be expected by anybody buying the NAS. The plastic wasn’t a deal-breaker for me, and I don’t think it should legitimately discourage anyone from buying a R1 Pro NAS. But if you hate plastic and you want to build a quality economical DIY NAS, you might want to avoid the R1 Pro NAS and check out last year’s economical DIY NAS blog instead.
Inside the Topton N1 NAS, there was only the fan on the CPU. I noticed that the drive bay opposite of the CPU was a little bit warmer than the other (~52° Celsius vs. ~43° Celsius). That warm–but not hot–drive was running at a higher temperature than any of the drives in my primary NAS, and I shared this concern when I reviewed the Topton N1 NAS.
In the Topton R1 Pro NAS, this inadequate cooling isn’t a problem anymore. They solved this by adding a fan in the bottom of the case that sucks cool air in and pushes it equally across both drives. During a strenuous replication task that was constantly writing data to the pool, the hard drives’ temperatures stayed around ~43° to ~46° degrees Celsius, which is just like the temperatures in my primary NAS.
In the product description, Topton gives a rather bleak warning about how you need to be a “professional technician” to install RAM and an SSD in the R1 Pro NAS. I don’t think this is true at all. It’s a simple matter of removing a few screws and withdrawing the inner frame. It was easy to dissassemble, easy to install the hardware, and easy to put it all back together. Mostly this warning exists to upsell other bundles which include RAM and SSDs.
Adding the hard drives to their sleds was quite easy, and they slid down nicely and gently. I really like the layout of the drives. Not just because it’s funny and the hard drives look a bit like chonky Pop Tarts going into a toaster, but it just makes more sense than trying to build something more cubish where the drives load in from the front.
I’m no “professional technician,” but I sure found putting together the barebones R1 Pro NAS to be rather easy.
Ever since its beta opened to the public, I’ve been an enthusiastic TrueNAS SCALE user. Debian Linux lies at the foundation of TrueNAS SCALE, and as a result, it opens the doors for a wider range of hardware support–particularly the consumer-level hardware.
The installation, setup, and configuration of TrueNAS SCALE was fairly unremarkable. Everything was set up and running just fine in a few minutes. I had a pair of Western Digital HC530 14TB hard drives which I created a 2-disk mirror out of.
For fun, I saturated the 2.5Gbps network using iperf3. And then I set up a TrueNAS replication task to replicate my media collection from my primary NAS and watched it whir along as fast as the mirrored disks would permit it to go.
No–but it almost could be! I am confident that if push came to shove, I could manage my data and squeeze it all onto 18TB HDDs (the largest supported according to the manufacturer), and I’m pretty confident that people will have success using hard drives larger than 18TB.
At its price, the R1 Pro NAS certainly seems like it’d be capable of meeting the penultimate goal for my EconoNAS builds, which I rarely accomplish: a fully functional NAS with redundant storage for under $500.
One of my complaints about my off-site backup NAS is that it lacks redundancy, and people caution against building ZFS pools out of multiple USB drives. I like the idea of some day building a new off-site backup NAS using the R1 Pro NAS.
I have two observations to share about the R1 Pro NAS, in no order of importance.
Firstly, there’s only a single SODIMM slot on this motherboard–as opposed to the two on the N1 I reviewed for Butter, What?!. I dug through pages of search results on Amazon for “64gb sodimm ddr4 1x64” and didn’t find a single listing. The 32GB of RAM I put in this R1 Pro NAS is likely the most it’ll ever support. 32GB of RAM is more than fine to be your NAS, but I anticipate that RAM would be limiting as you explored the world of Virtual Machines and Apps under TrueNAS SCALE.
Secondly, other than swapping out smaller hard drives for larger ones, you’re never upgrading this DIY NAS. One of my favorite things about my regular DIY NAS builds is that their standardized form factors mean that there’s a possibility of upgrades down the road without starting over from scratch.
Well, I’m selling them–so my answer is going to be a resounding “Yes!” But I’m also acting as a salesman here, so take that with a hearty grain of salt!
Whether you should buy it depends on whether it will meet your needs. Primarily, I think you should wonder about its storage capacity, your need for redundancy in your hardware, and how quickly you are hoarding data.
I really like that the R1 Pro NAS is small, it can accomodate high-capacity hard drives, its CPU sips power–but you get a big bang for the buck, it comes mostly assembled, it runs TrueNAS SCALE, and it’s quite affordable.
If you’ve been thinking about building a NAS and you’re worried about breaking the bank, worried about having to dedicate a bunch of room in your home to another computer, or worried about the potential cost of building a “serious” DIY NAS, then I think you should be seriously considering the Topton R1 Pro NAS.
I liked the Topton R1 NAS Pro enough that I emptied my blog’s bank account to buy as many as I could afford–and I’ll definitely be holding on to one for myself!
]]>Technically, the answer to this question is yes. Jekyll is at the foundation of Octopress. But the way Octopress 2.0 was constructed, upgrading the version of Jekyll that it was using was somewhere between impossible and difficult.
Unfortunately, at this point Octopress is derelict and appears to be abandoned. Octopress’s creator teased that Octopress 3.0 was on the verge of being released back in 2015. But unfortunately, that release never materialized, and there hasn’t been any subsequent news from Octopress since. In the past ten years, I’ve rebuilt my desktop computer a couple times and learned that some of the Windows prerequisites were so old they couldn’t be installed anymore, and the current versions weren’t backwards compatible. After my most recent PC upgrade, I couldn’t cleanly recreate my Octopress environment. I spent a few days fiddling with it and got it working with newer–but not current–versions of Octopress’ dependencies, but I’ve been worried about whether I did a good job or not. It simply became obvious that moving away from Octopress was inevitable.
To make matters worse, Google hasn’t been as kind to my blog as it used to be. Broadly speaking, I think this applies to a lot of blogs–not just mine. I still rank really high on the same search terms that I have always ranked high on, but for one reason or another dramatically less traffic is sent my way by Google. The overwhelming majority of my traffic finds its way to me via Google, so this has led to a significant decline in visitors over the years.
I don’t believe that the ancient static site generator is to blame for this drop in Google traffic, but it’s indisputable that I was missing out on more than a decade’s worth of development to Jekyll. I am optimistic that in these past ten years, there have been some improvements to Jekyll that will hopefully lead to an increase in traffic to the site.
I’ve created quite a bit of content in the time that I’ve been using Octopress, and in that time, generating my site has gone from something that used to take a few seconds to something that took quite a few minutes. Generating and deploying my site has been taking way longer than I would’ve liked. How long it takes to generate and deploy both of our blogs became painfully obvious to Pat and I after we launched butterwhat.com. When working on content for Butter, What?!, it takes a moment to acclimate back to how slow it is to when working with briancmoses.com.
I took a look at a few different static site generators but quickly decided that my easiest path was to stick with Jekyll and upgrade to a current version. But between changes to kramdown, the Jekyll Origin theme from zerostatic, upgrades that I wanted to make to my existing blogs, and the consequences of my own shenanigans, I wound up having to update each and every one of my blogs. All by itself, it took me a couple weeks to make all these changes! Beyond that, I spent some time tinkering with other Jekyll add-ins and customizations to the theme to make sure everything worked as well–hopefully better–as they did on my old site!
Literally everything has changed, but I thought I’d list a few of my favorite changes:
I actually deployed all these changes at the very end of October. I’ve been anxiously monitoring the site, looking for problems that thankfully have not popped up! I wanted to wait a couple weeks and let it bake in before I started writing this blog. So far, I’ve been encouraged by the feedback that I’ve gotten from the behind-the-scenes content that I’ve shared in the Butter, What?! Discord server and on my Patreon page.
If you’ve got any thoughts on the update to my website, share them in the comments! I’d love the feedback. If you found something broken, please contact me! An embarassing enough mistake surely is worth a coupon at my Tindie store or eBay store!
]]>Nearly every year, the EconoNAS lacks two killer features that an off-the-shelf NAS always has: diminutive footprints and hot swap hard drive bays. This year, I felt ambitious and tried to close those gaps by setting a few additional goals:
These three new objectives made this year’s EconoNAS an even bigger challenge! There are significant price premiums for small form factor PC builds and for cases which feature external drive bays.
For the past few years, I have not been buying the parts and building the EconoNAS myself.~ This year’s EconoNAS is no different. I always make every effort to try and pick components that would work well together, but it’s possible that some sort of incompatibility is lurking out there that I’m not aware of or have overlooked.If you wind up building the DIY NAS: EconoNAS 2023 (or something similar) for yourself, then please share your experience in the comments!
Update (10/01/2023): On Twitter, @TrueNAS challenged me to build and give away the DIY NAS: EconoNAS 2023! I was eager to accept and complete this challenge. The last component arrived during the day on Friday (9/29) and by Sunday evening (10/1) I had compelted assembling the EconoNAS. 20 days from this update I’ll complete this challenge by selecting the winner of the giveaway!
When I built the DIY NAS: 2023 Edition, I was astonished at what a good value it wound up being. In fact, it was such an excellent value that I nearly decided to skip this year’s EconoNAS altogether. Rather than do that, I figured that I would make the Topton N5105 NAS Motherboard the centerpiece of this year’s EconoNAS, too. The motherboard features:
Given that I didn’t actually build this year’s EconoNAS, picking the Topton N5105 NAS motherboard is a no-brainer. I’m already very confident in this motherboard’s ability to be a DIY NAS as well as comfortable with its compatibility with TrueNAS SCALE.
Picking out a case is going to make or break this year’s goals for the EconoNAS. Historically, small form factor cases with external drive bays have always been expensive, especially when they’re Mini-ITX.
I was really excited when I discovered the AUDHEID K3. It’s a compact case with four external drive bays which seems ideally suited for a motherboard like the Topton N5105 NAS motherboard.
I only wish that the AUDHEID K3 case had some internal 2.5” drive bays so that one could fully utilize all the motherboard’s storage capacity. If I were building this EconoNAS, I would be looking for gaps in the case where I could modify the case to squeeze in a pair of 2.5” SATA SSDs. For example, I could create the gap by moving the case’s cooling fan to the outside of the case. This would be a fun component to 3D-design and print on my 3D-printer. Alternatively, I would consider using Velcro or double-sided tape to adhere the SATA SSDs into any gaps inside the case.
When I picked out the components for this EconoNAS, I knew that I would need to save money on certain components. I was quite tempted by picking the bare-minimum amount of RAM recommended for TrueNAS SCALE, 8GB. However, I opted to go with this 16GB kit of 3200MHZ DDR4 RAM from Crucial instead.
I currently run 8GB of RAM in my off-site NAS, and in the few months that I’ve been using it, I think it has done a fine job. The biggest driver in the EconoNAS is extracting the most value out of the dollars invested. I simply think there’s even more value in having 16GB of RAM that it justifies the extra $20 that it costs. If your budget allows it, upgrading to 32GB or 64GB of RAM are both good values too!
It is wise to divvy up storage hardware into two categories: boot and user storage. Drives purchased for the boot drives in the EconoNAS are usually low-capacity SSDs with little to no hardware redundancy. Drives purchased for user storage are higher capacity, can be HDDs or SSDs, and hopefully there’s one–or more–drives’ worth of hardware redundancy in that user storage.
For all of my DIY NAS builds, I aim to spend less than $20 per boot drive. Because I prefer the TrueNAS appliance OS (SCALE or CORE), it is important to understand that all of the boot drives’ capacity is dedicated to the appliance and that none of the capacity will be accessible for the purpose of user storage. TrueNAS doesn’t require very much capacity at all (quite a bit less than 16GB), which makes finding SSD in this budget pretty easy.
For this year’s EconoNAS, I picked out the Silicon Power P34A60 128GB NVMe (specs). At 128GB, it has eight times the capacity that it needs to have. Drives with a smaller capacity exist, but if they’re from a company that I recognize, they’re usually as expensive–or sometimes more expensive!
The costliest part of any DIY NAS build is ultimately going to be the drives used for storage. The amount of money spent on storage increases with capacity, redundancy, and the performance of the storage. What kind of storage and how much of it is very specific to the person who is building it.
The focus of the EconoNAS is value, which I measure using price-per-terabyte, but the best value on storage is almost always the largest drives which also have the biggest price tags. And the drives with lower price tags are usually a pretty bad value.
I don’t have any current recommendations for storage to buy, but I have a few tips to share for getting the most storage for as little money as possible and I’ve included a list of hard drive deals that I update from time to time:
Always avoid shingled magnetic recording (SMR) drives when working with the ZFS filesystem!
Hard Drive | Price | |
---|---|---|
HGST Ultrastar He10 (HUH721010ALE600) 10TB (Renewed) | specs | $77.50 $7.75/TB |
Western Digital Ultrastar DC HC520 (HUH721212ALE600 ) 12TB (Renewed) | specs | $118.99 $9.92/TB |
2-pack Seagate Exos X16 16TB ((ST16000NM001G) | specs | $433.99 $13.56/TB |
Each year, when shopping for the EconoNAS’ power supply, I generally buy the cheapest power supply that I can find that fits the power needs of the rest of the components. When I was considering the AUDHEID K3, I was very worried that I wouldn’t be able to find an affordable FlexATX power supply. To my surprise, the Apevia ITX-AP300W (specs) came in at a very reasonable price.
Generally speaking, I think one of the best PC upgrades is to put in a high-quality power supply. If there’s room in your budget, then the power supply is a component that I would certainly recommend upgrading in this build. Here are a few alternatives:
Back in 2022, I rebuilt my own DIY NAS around TrueNAS SCALE once it hit its first beta release. For the most part, I think using TrueNAS SCALE is incredibly similar to TrueNAS CORE (formerly known as FreeNAS). But there is one area where TrueNAS SCALE really sets itself apart from TrueNAS CORE: hardware compatibility. TrueNAS SCALE is built upon Debian Linux, and TrueNAS CORE is built atop FreeBSD. Very broadly speaking, hardware support is simply much better under Linux than it is in FreeBSD.
Beyond that, TrueNAS SCALE’s other unique features (the KVM hypervisor, Linux Containers, etc.) seem to be superior to TrueNAS CORE’s equivalent features. Lastly, I feel like TrueNAS SCALE is more well suited for the storage enthusiast who wants to better utilize their hardware by turning their NAS into more of a homelab machine.
Component | Part Name | Qty | Cost | |
---|---|---|---|---|
Motherboard | Topton N5105 NAS Motherboard | specs | 1 | $199.99 |
CPU | Intel Celeron N5105 | specs | N/A | N/A |
Memory | Crucial RAM 16GB Kit (2x8GB) DDR4 3200MHz | specs | 1 | $40.46 |
Case | AUDHEID K3 | N/A | 1 | $79.99 |
Power Supply | Apevia PS-ITX300W Flex ATX 300W Power Supply | specs | 1 | $34.99 |
Boot Drive | Silicon Power 128GB M.2 NVMe SSD | specs | 2 | $25.97 |
TOTAL: | $378.40 |
Within moments of accepting @TrueNAS’ challenge on Twitter I placed an order for the parts that I needed to build the DIY NAS: EconoNAS 2023. In the following days those parts slowly started arriving and I burned in the motherboard and RAM. I then installed the NVMe SSD and loaded it up with TrueNAS SCALE. This past Sunday (October 1st) I assembled the EconoNAS and filled up each of its drive bays with 14TB hard drives that I had on hand for a future project.
For the most part, I didn’t have any major issues assembling the EconoNAS. Everything fit together and it was a completely functional NAS shortly after creating a new pool and dataset after its first boot. If you’re planning on building the DIY NAS: EconoNAS 2023 for yourself, I ran into a few minor challenges that I felt were worth sharing:
I’m glad I got the opportunity to put together the DIY NAS: EconoNAS 2023
I spent a little bit of time digging around the current NAS products from QNAP, Synology, Terramaster, and Asustor, searching for a comparable NAS with four drive bays and an x86 CPU. For each manufacturer, I wound up picking the NAS that I felt was most equivalent to DIY NAS: EconoNAS 2023. And here’s what I found:
Each of these off-the-shelf NAS products is more expensive than the DIY NAS: EconoNAS 2023. With regards to the CPU, all but the Asustor AS5404T have weaker CPUs, and the AS5404T has the same CPU but costs nearly $150 more than the EconoNAS. All four of these NAS appliances only have 4-8GB of RAM compared to the 16GB of RAM in the EconoNAS
However, it’s important to remember that each of these off-the-shelf NAS machines is already assembled and has some sort of a customer support department. Not having to assemble the NAS and not having to support it yourself are each compelling features, but they’re just not features that I value very much. It is important to realize and acknowledge that value is very subjective and varies widely by person. If these features are something that you value, then I liked the Asustor AS5404T best out of this group.
Ever since my very first EconoNAS build, I’ve been wishing that I would be able to build an affordable, small form factor NAS. This is something I’ve tried to do each time I pick out parts for an EconoNAS, but it always winds up being too expensive. That pattern broke this year, and I managed to accomplish a number of firsts for the EconoNAS build:
The first time I added up the cost to all of the parts, I was astonished that it was only about fifteen dollars more expensive than the DIY NAS: 2022 EconoNAS. I was so surprised, I had to triple-check my parts list for omissions and quadruple-check my math! I expected a NAS with this many features would’ve been considerably more expensive than last year’s EconoNAS.
Every year, I expect that each NAS will be better than the prior years’, but with the EconoNAS, I’ve often found that they all kind of blur together. Usually its a big black ATX case, a closeout-priced motherboard, and an inexpensive CPU with an integrated GPU.
But this year turned out different! This EconoNAS is hands down the best that I’ve ever put down on paper. What do you all think? Let me know in the comments!
I really like that both the DIY NAS: 2023 Edition and the DIY NAS: EconoNAS 2023 both use the Topton N5105 NAS motherboard. I’m excited because I have quite a bit of faith from my firsthand experience with the motherboard using it in the earlier DIY NAS build. This helps me be quite a bit more confident in this build, despite the fact that I did not build this NAS. I’m also hopeful that people who are reading this blog a few months down the road will have a greater opportunity to still be able to obtain the motherboard.
I’m also excited that in this EconoNAS build, headroom exists for upgrades to the RAM, an empty M.2 slot, two empty SATA slots (but no obvious place to mount drives), and if you really wanted a CPU upgrade, then a variation of this motherboard exists with a Pentium Silver N6005 CPU:
What do you all think? Does unlocking the Mini-ITX and hot swap bay impress you in a budget-focused DIY NAS? Do you like this year’s EconoNAS? Would you prefer the DIY NAS: 2023 Edition instead? Or would you build something entirely different and uniquely suited for your needs? Let me know in the comments and come join the #diynas-and-homelab channel in the Butter, What?! Discord server to tell us all about your DIY NAS!
Within moments of publishing this blog, the TrueNAS Twitter account had this to say:
We DARE @briancmoses to build this and give it away to the community. We will even throw in $100 of #TrueNAS Merch to the winner of the giveaway. https://t.co/HvR9vMPgLo
— TrueNAS Open Storage (@TrueNAS) September 21, 2023
Naturally my reply was to accept TrueNAS’ challenge! There are 7 total ways to enter the DIY NAS: EconoNAS 2023 featuring TrueNAS. The contest ends on October 21st and the winner will be drawn and contacted. In the event that the winner does not respond within 72 hours a new winner will be drawn, this cycle will repeated until a winner responds!
While everyone is entering the contest, I will be building, burning in, and doing some testing of the DIY NAS: EconoNAS 2023. Good luck to us all!
]]>Over the years, I’ve slowly broadened the amount of my data that I’m backing up to Backblaze B2, and thanks to my YouTube channel, the amount of data that I’m accumulating has increased quite a bit in recent years. When my monthly charges eclipsed $50 a month, I felt like it was time to start thinking of alternatives.
Realizing that it was only inevitable that my costs would continue to climb, instead of continuing to pay every month for Backblaze B2, I figured that I’d build an inexpensive NAS and use the buddy system to host it off site.
An off-site backup is a copy of your data stored at a location away from the rest of your data. Off-site backups are an important part of the 3-2-1 backup strategy because an off-site backup helps protect you against theft, household disasters, and—depending on their magnitude—natural disasters.
The buddy system is pretty much exactly what it sounds like: work with a buddy—a buddy that you trust—and each of you hosts the other’s backups in your respective home.
For a little over a year and a half, I’ve been hosting Pat’s off-site storage server here in my office. Pat’s off-site NAS is a Raspberry Pi and a USB hard drive running Seafile and Tailscale. Pat uses this to keep his data synchronized across several devices, including revision history. I’ve been self-hosting my own cloud storage using Nextcloud, which would’ve made it simple to emulate Pat’s solution. I could have easily substituted Nextcloud in for Seafile and moved most of my data into Nextcloud, but I had something else in mind.
As a TrueNAS enthusiast, I really wanted to leverage TrueNAS as part of my solution. I’ve been really happy using TrueNAS’ Cloud Sync Tasks to back up my data to Backblaze B2, and I wanted to use TrueNAS’ Replication Tasks to accomplish something very similar to my off-site NAS. TrueNAS’ hardware requirements and the Raspberry Pi foundation’s decision to abandon the consumer market amidst their supply-chain issues meant I wouldn’t be using a similar hardware setup to Pat’s.
But before I dive into my solution, it’s important to talk about the most important aspect of all.
In order for any kind of buddy system hosting to really work, you have to trust your buddy, and you have to be trustworthy! There’s a lot you can do to protect a machine that you’re leaving at a friend’s house, but there’s only so much that you can do when your buddy has physical access to your hardware.
There are all sorts of nefarious things that Pat could do because he has physical access to this NAS. There are all sorts of nefarious things that I could potentially do with this machine on Pat’s network. The buddy system works for us because we’ve built that trust and put in the work to maintain it.
For my off-site NAS, I decided that I’d use a Mini PC. I’ve been interested in the Intel NUC, the Mac Mini, and many other Mini PCs for a long time, but I’ve never had a reason to buy one. Before Pat first mentioned his idea for his off-site file server, I never would have imagined building a NAS with a Mini PC.
I ultimately decided to buy the Beelink Mini S (specs). Its specifications–an Intel Celeron N5095 CPU, 8GB of DDR4 RAM, and 256GB SSD–help it meet the bare minimum hardware requirements for TrueNAS SCALE. My experience with the DIY NAS: 2023 Edition had me confident that the Celeron N5095 CPU was more than ready to handle the tasks I needed it to perform using TrueNAS SCALE.
Beelink has a fantastic range of other Mini PCs too. Folks in our Discord server have shared their success stories doing all sorts of interesting NAS-related or homelab-related activities. I recently recommended a Beelink SER5 to my parents to replace their ancient desktop PC and have considered something similar to replace my wife’s old desktop computer too.
If I were starting this project over today, I probably would’ve spent the few extra dollars to get the Mini S12 Pro, and if I needed more compute power, I wouldn’t have hesitated on purchasing the SER5 Pro, SER5 Max, or any of the rest of the Beelink Mini PC lineup.
Finding storage was a bit of bigger challenge! My Backblaze B2 buckets had reached 10TB of capacity. I wanted my off-site NAS to be able to back up everything that I’d been sending to Backblaze B2, I wanted to institute some rudimentary snapshots, and I hoped I’d have a few terabytes left over for the future.
In the beginning of my search, I found that the largest of external hard drives were barely large enough, and the best deals that I was finding usually were for drives that didn’t have quite enough capacity. So I patiently waited to pounce on a great deal. I’ve been hyper-focused on an awesome deal for a 18TB (or larger) USB hard drive. Months later when I noticed that the WD Elements 20TB USB hard drive went on sale for $279.99 at the end of March, I sprang into action and bought the hard drive to use in my off-site NAS.
If you’re interested in deals like this, I regularly share these kinds of deals in the #deals channel of the Butter, What?! Discord server whenever I come across them. In 2023 alone, I’ve purchased 76TB worth of hard drives from deals that I’ve seen or posted in our Discord server!
Component | Product | Qty | Cost | |
---|---|---|---|---|
Mini PC | Beelink Mini S | specs | 1 | $139.99 |
Hard Drive | Western Digital 20TB Elements | specs | 1 | $279.99 |
Total | $419.98 |
In order to accomplish what I set out to do, I needed both Tailscale and TrueNAS SCALE. Of the two, Tailscale was more important, as it is my VPN software of choice. TrueNAS makes using Tailscale considerably more difficult than it is to use on literally any other platform that I use, but thankfully it was, not impossible.
I wound up using the Tailscale app from the Truecharts catalog and mostly following Truecharts’ how-to guide. But because TrueNAS’ replication tasks use SSH, I had to veer from the guide and make these changes:
Host Networking
(under Expert - Pod Options
)Userspace
(under App Configuration
)Once my new off site NAS was added to my Tailnet, I also made sure to disable Tailscale’s key expiry for the new machine. As Bruce D. points out in the comments, should this node’s Tailscale key expire in the future then I’d definitely need Pat’s assistance or I’d have to go visit his house in order to obtain a new key for my remote NAS,
I followed the TrueNAS SCALE guides on replication tasks and had my primary NAS replicating all of its contents over to this secondary NAS in no time. Because of the amount of data involved, I let this initial replication complete on my local network.
In order to have confidence that the replication would continue to work once this secondary NAS was off my network, I tethered my GL.iNet Slate travel router to my mobile phone, plugged the secondary NAS into the travel router, and reconfigured the NAS network settings for the new router.
The next 2–3 days I let my replication tasks run overnight with the secondary NAS connected to a completely different network. Everything worked just fine until my I ran out of my allotment of mobile hotspot minutes from T-Mobile. The tasks continued to work just fine, but they were so badly throttled that it didn’t make any sense to continue.
Pat invited us over for pizza recently, and I saw it as an opportunity to get my new NAS deployed, although it felt a tiny bit rushed. I would have liked to have seen it in action at home, but communicating via Tailscale, for another few days before dropping the NAS off.
Regardless, I packed it up, enjoyed some pizza, and hurriedly plugged it in to Pat’s network at the end of the night. Within a couple minutes it was connected to my Tailnet, I’d logged in to the TrueNAS SCALE interface using the machine’s Tailscale IP, and I kicked off one of my replication tasks on my primary NAS. A few moments later, that replication task had succeeded, and I was confident in my off-site NAS’ ability to perform its function.
I’m simultaneously running TrueNAS cloud sync tasks to Backblaze B2 and replication tasks to my off-site NAS. Before the end of my next Backblaze billing cycle, I anticipate that I will be able to delete my TrueNAS cloud sync tasks along with my Backblaze B2 buckets.
My monthly bill to Backblaze B2 has hovered around $60 for the last year. After spending $420 to build my 20TB off-site NAS, I expect to break even in about 7 months. It’s worth pointing out that there’s some risk here–the hardware could fail before it pays for itself. The products’ warranties might be helpful here, but the warranty process might be long enough that I’d just go ahead and buy a new set of hardware first!
What should I wind up spending that extra money on? I’ve got a few ideas:
In my shoes, which option would you pick? Let me know down in the comments, or better yet, come tell us about it in the #diynas-and-homelab channel of the Butter, What?! Discord server!
]]>I changed things up a bit this year and had the contestants publicly share their answer to this question along with a link to the DIY NAS: 2023 Edition:
If you won the DIY NAS: 2023 Edition, what would you do with it? What sort of problems would it solve for you?
I have genuinely enjoyed reading each and every one of the entries to the giveaway. It has been equally entertaining and interesting to see how others would put the DIY NAS: 2023 Edition to use.
I enjoyed reviewing the entries so much that I wanted to make sure this announcement included not only the winner, but a handful of my other favorite entries too.
Each of these honorable mentions deserve being mentioned in this blog. While they fell just short of winning the contest, each one of them was compelling on their own. Having an abundance of great entries was a blessing, but it certainly made picking a winner a bigger challenge!
When I first learned of the Topton N5105 NAS Motherboard, I immediately shared it with Pat. At the time, I had not picked any parts for the DIY NAS: 2023 Edition. But in just chatting about the motherboard with Pat the next few days, it became obvious to me that it was the best option for what I wanted to do with this year’s DIY NAS. Pat wound up writing two of his own blogs; first an entry into the contest and second about what a nice pairing the Topton N5105 NAS Motherboard and JONSBO N1 make for a DIY NAS.
In his entry, Pat wrote:
I know that Brian would agree with me that it wouldn’t be fair for me to win the NAS.
Pat’s correct; it would seem a bit unfair if he were to win the contest. Nevertheless, either of Pat’s two blogs are fantastic entries into the contest and very well could have won him the motherboard. If you’re considering building the DIY NAS: 2023 Edition, Pat’s two blogs are must-read!
I guess now I need to worry about Pat following through with his threat of sneaking off with one of the Topton N5105 NAS motherboards next time he’s over at my house!
… Tyler S., for his blog, “What I would do if I won a NAS?” In the blog, Tyler talks a laundry list of things that anyone could do with a DIY NAS: back up critical data, create shared folders that can be accessed from other computers, set up a media server, host your own game server, and Tyler talks about several other homelab activities. What I really liked about Tyler’s entry is how he seems to be the spokesperson for anyone who doesn’t currently have a NAS, not just himself. Tyler listed off many problems that he’d solve with the DIY NAS: 2023 Edition, but he answered them in a way that applies to the reader too. If you’ve been wondering if you need a NAS or not, Tyler’s entry into this year’s giveaway is a fantastic resource!
Thanks again to everyone who entered! I appreciate each and every one of the entries. Thanks to the change in the giveaway format, there were far fewer entries than prior giveaways. But the quality of each entry was so much higher! I enjoyed reading each entry, sharing it on social media, and then poring over each one to pick a winner!
What’d you think about the giveaway of the DIY NAS: 2023 Edition? If you missed out on the giveaway, are you determined to build a DIY NAS of your own? Let me know in the comments, or come chat with us in the Butter, What?! Discord server.
]]>Ever since Pat decided to buy the Prusa I3 MK3S back in 2020, I’ve been wondering if I should buy a new 3D printer. This wonderment has been exacerbated a few times when I was bottlenecked by fulfilling orders from my Tindie store.
All the stars seemed to align back in November of 2021 when Prusa announced the Prusa XL. The Prusa XL was supposed to be bigger, faster, and way more advanced than my Prusa MK3. The announcement claimed printers would start shipping in “Q2/Q3 of 2022.”
The day of that announcement, I paid the deposit, and then I began waiting for the preorders to begin getting fulfilled. In the time that I spent waiting, a few other things happened:
As of the writing of this blog, Prusa is still not shipping what I wanted: the multi-tool version of the Prusa XL.
Once Prusa let me know that the earliest I’d see my printer was May of 2023, alternatives quickly came into focus. I could buy the Bambu Lab X1-Carbon Combo ($1,450) and an AMD Radeon RX 7900XTX GPU ($999) for far less than the 5-tool Prusa XL ($3,500) that I planned to buy. Even better, there wouldn’t be a risk of the eventual delays that have followed.
When I realized I could have nearly everything that I wanted, that I wouldn’t have to worry about more delays, and that I could save $1,000 in the process, I canceled my preorder for the Prusa XL and immediately bought the Bambu Lab X1-Carbon Combo and an AMD Radeon RX 7900XTX GPU instead.
As of the writing of this sentence, I’ve logged hundreds of hours of print time on my Bambu Lab X1C in the months following its delivery. Getting the printer unboxed, set up, updated, and printing took less than an hour. Most of those hours of print time have been 3D-printing things for my new 3D printer, but I have fulfilled a few Tindie orders of my ESP32 D1 Mini cases too. If these first 175 hours are any indication, I’m quite impressed!
It took nearly all Sunday to #3dprint this little Spider-Man figurine for my son. I’m impressed with how it turned out and how well the Bambu Labs X1C churned through it. pic.twitter.com/mCWztN9RnV
— Brian Moses (@briancmoses) April 17, 2023
As much as I like the X1,C there are a few things about it that worry me a bit.
While my new printer was in transit, Prusa announced the release of the Prusa MK4, and I said to myself, “Oh no, now it is time for my Buyer’s Remorse to kick in!”
I immediately began to worry that the Prusa MK4 might have been a better printer for me. I’m no stranger to Buyer’s Remorse, and thanks to a series of concerning missteps, Prusa Research helped me realize that the Bambu Lab X1-Carbon was the ideal printer for me.
Enough of the product page for the MK4 printer or its kit(s) is fictional that I’m as alarmed as I am disappointed. If you were to buy the MK4 today, you wouldn’t get input shaping—so you’re definitely not printing a Benchy in under 20 minutes that they’re bragging about. If Josef Prusa’s blog is any indicator the MK4 may not be an actual “open-source” printer, you aren’t getting the 10% off you were led to believe you’d get by buying previous i3 printers, and unless you ordered the MK4 quickly (when it’s back-ordered for 6–7 weeks) you didn’t receive the 1Kg of filament that was promised to buyers when it was released.
A tiny bit of proofreading, a newfound commitment to honoring promises made to its customers, and a firmware release might be all it takes to break this recent pattern. But until that pattern is broken, I’ll continue to be glad that I purchased the Bambu Lab X1-Carbon Combo.
Heck no! I’m flustered and discouraged by what I’ve seen recently from Prusa Research, but that doesn’t completely erode goodwill accumulated from years of hassle-free printing!
My plan all along was to give me the option of simultaneously restocking my Tindie store while enabling me to work on other 3D-printing projects. A second printer has always been my plan, and I do love when a plan comes together!
Over the past two months, I’ve been a 3D-printing maniac. I’ve been really enjoying working with the Bambu Lab X1-Carbon. I’d probably be printing this much regardless of what 3D printer I bought. But maybe I wouldn’t have printed this much stuff:
Two weeks from when this blog is published, I won’t be surprised one bit if I’ve managed to eclipse 200 hours on my X1C. I already have a few projects in mind for my office, for my Tindie store, and I’m half-tempted to print a whole new MK735 case for my DIY NAS.
I’m genuinely excited to 3D-design something and crank out iterations until I’m pleased with how it turned out. I’m ecstatic that I won’t feel have to choose to either print something for fun or restock my Tindie store.
Are you looking to add a second 3D printer to your setup? I certainly think that either the Bambu Lab X1C or P1P are fantastic values. But I also think you should check out the Sovol SV06 like Pat did. Much of the value is just having a second printer!
Are you thinking about buying a Bambu Lab X1-Carbon, a Bambu Lab P1P, a Prusa MK4, a Prusa XL, or something else? Come join us in the #3dprinters-lasers-cncs channel in The Butter, What?! Discord server and tell us all about it!
I scaled this Venom #mini13 figurine up to 400%! It's huge, I included a 12 ounce beer for scale. I hope my son likes it! pic.twitter.com/acugAxLCmR
— Brian Moses (@briancmoses) April 19, 2023
I like to think that my key criteria for building a DIY NAS build has been pretty consistent over the years; that criteria is:
My desire for a diminutive, quiet, and power-efficient DIY NAS was born from a lack of space in my office from ten years ago. My office has grown considerably since then, but I still value these features and strive to meet them with each of my DIY NAS builds.
The minute I learned about the Topton N5105 NAS Motherboard, I knew that it was going to be in the DIY NAS: 2023 Edition. In fact, I was so excited about it that I ordered it the very next day after learning about it. When I couldn’t find any US-based vendors, I was inspired to buy some from the manufacturer and to open a store on eBay and see if I could sell some myself!
The motherboard features:
Nearly every one of my ideal criteria for a DIY NAS is checked off by this incredibly interesting motherboard. It supports up to 8 drives (2x M.2 and 6x SATA), it fits a small form factor, and its CPU sips power. The only criteria that it does not meet is that the CPU requires a fan. The icing on the cake on the Topton N5105 NAS Motherboard is that it includes four 2.5Gbps network interfaces. For what you’re getting, I believe that the Topton N5105 NAS Motherboard is a fantastic value to use in a DIY NAS build.
One of my bigger gripes every year is that there are just not very many NAS-specific cases being designed each year. I strive to pick a new NAS-centric case each year, but each year the landscape of NAS cases is pretty static. In March of 2022, I caught a glimpse of the JONSBO N1 on the Linus Tech Tips YouTube channel. The release of the JONSBO N1 (specs) solved a big problem for me in 2023.
The JONSBO N1 has room for a MiniITX motherboard, requires a SFX power supply, has room for up to five 3.5” HDDs, and room for an additional 2.5” HDD. Personally, I’d like to see room for at least one more 3.5” HDD but I’m not deducting points for that. My criteria of six 3.5” HDDs was established back when you were lucky to get a 2TB HDD at $200. In 2023, we’re regularly seeing hard drives five times as big (or bigger) under $200. I believe it might be time to rethink this particular criteria. Unless you’re a serious data hoarder, five drives in 2023 can be more than plenty!
I spent most of 2022 being excited about building a DIY NAS inside of the JONSBO N1, and that anticipation paid off at the end of the year!
The Topton N5105 NAS motherboard has two laptop-style DDR4 SO-DIMM slots, so I opted to max it out with a Crucial 32GB kit of DDR4 2933Mhz RAM (specs). One of my only disappointments with the Topton N5105 NAS motherboard is that its capacity for RAM tops out at 32GB. Had I realized that the Topton N5105 NAS motherboard actually supported up to 64GB of RAM, I would’ve gladly bought twice the RAM for twice the price.
For the ECC zealots out there, I agree ECC RAM is a better choice–but that doesn’t mean it is a requirement. I tackled this quite a few years ago when I shared why I chose Non-ECC RAM for my own DIY NAS back in 2012. In the years since, I’ve built up and replaced all of that hardware a couple different times, and I happen to be using ECC RAM today.
However, I stand by what I wrote back in 2014. For the DIY NAS: 2023 Edition, the motherboard, CPU, and RAM added up to just over $350. I’m reasonably confident that an equivalent motherboard, CPU, and ECC RAM would wind up being twice as expensive. Even if you wanted to spend that extra money, I think there’s more value in spending it differently than there is in spending it on ECC RAM.
For the past few years, I’ve divvied the storage devices into two categories: storage for the boot drive and the devices for the storage of your data.
In last year’s DIY NAS build, that line got a bit blurry as I partitioned the SSDs and used a slice of them for the boot device and then used the leftover space as storage for my VMs and containers. While I’m happy with how this has worked out, I’m not necessarily certain if I’d encourage others to do the same. It’s not a supported solution and requires going in and modifying the installation script prior to installing TrueNAS.
I think that the most debatable choice that I made was picking a boot device. In the JONSBO N1 there is room for five 3.5” HDDs and one 2.5” HDD. The Topton N5105 NAS motherboard supports up to 6 SATA devices and two M.2 devices.
Because I like having redundancy on the boot drive, I opted to pick up a pair of 250GB PNY CS1030 NVMe SSDs (specs). 250 GB of capacity is definitely overkill for the capacity, but at around $20, the price was in the right ballpark.
If maximizing data density were my goal, I probably would’ve picked a single SATA SSD like the 240GB Crucial BX500. That would create the possibility of using the motherboard’s two M.2 slots for a pair of NVMe SSDs for some additional storage capacity to the NAS.
For the sake of writing this blog, I pulled out five of my retired 4TB HDDs and loaded them into the DIY NAS: 2023 Edition. These are just excess drives that have been either upgraded or removed because their SMART test results have been suspicious.
It has been a while since I actually bought storage for one of DIY NAS builds. Most of the money spent on a NAS will be spent on hard drives, and the amount of capacity you’ll need depends on how much data you’re hoarding and how quickly you’re adding to it.
If I were building this for my own use, I think I’d be pretty interested in mirrored pairs of higher-capacity drives. I’d leave the fifth drive bay empty and use a spare hard drive to sneakernet large chunks of data between machines. For my needs, hard drive capacity has steadily outpaced how quickly I’m accumulating data. I no longer need half a dozen drives to hold all of my data like I did when I got started.
When I’m shopping for hard disk drives, I usually get pretty excited when I see a drive with a price-per-terabyte below $15. I very frequently see 12-16TB shuckable external hard drives hit this price. This Toshiba MG08ACA16TE 16TB 7200RPM SATA Enterprise HDD is on the verge of being tempting at $~260 ($16.25 per TB).
We are constantly sharing deals on hard disk drives in the #deals channel of the Butter, What?! Discord server. If you’re looking for a good deal or wanting to share a good deal that you have found, please join!
Hard Drive | Price | |
---|---|---|
HGST Ultrastar He10 (HUH721010ALE600) 10TB (Renewed) | specs | $77.50 $7.75/TB |
Western Digital Ultrastar DC HC520 (HUH721212ALE600 ) 12TB (Renewed) | specs | $118.99 $9.92/TB |
2-pack Seagate Exos X16 16TB ((ST16000NM001G) | specs | $433.99 $13.56/TB |
Motherboard manufacturers have been cutting costs by skimping on things like SATA cables for years. But with the Topton N5105 NAS motherboard, they’ve taken it to a new level!
The motherboard didn’t come with any SATA cables whatsoever. In this scenario, this is actually preferable because the amount of room in the JONSBO N1 case requires that you use right-angle SATA cables.
It is very common for small form factor cases to require a small form factor (SFX) power supply too. Being somebody who’s a big enthusiast for small form factor DIY NAS builds, I understand and accept this reality–but I kind of wish that weren’t the case. I’d gladly trade a bit of case volume for a wider choice of compatible power supplies.
The JONSBO N1 is no different than many of the other Mini-ITX cases I’ve used in the past. A SFX power supply is required. As a result, I picked the SilverStone Technology 450W SFX (ST45SF-V3-USA) power supply (specs).
In picking out the power supply, I did some back-of-the-napkin math and attributed about 10–20W to the Celeron N5105 CPU, 5–10W for the each of NVMe SSDs, and up to 25W for each of the five 3.5” HDDs. I probably would have been fine with a 300W power supply, but I opted for the 450W power supply to be safe.
My biggest concern in building the DIY NAS: 2023 Edition was that I’d discover that Linux support of the hardware found on the Topton N5105 NAS motherboard would be lacking.
The Topton N5105 NAS motherboard was enough of a value that I gambled the hardware would be supported by TrueNAS SCALE, and thankfully, the gamble paid off. But had it gone a different direction, I would’ve gladly switched to another NAS appliance like UNRAID, OpenMediaVault, or building a NAS from the ground up using a recent flavor of Linux.
It’d be overly melodramatic to say that I breathed a big sigh of relief after I installed TrueNAS SCALE without any issues and had the NAS up and running, but I was definitely relieved.
In my opinion TrueNAS SCALE is a fantastic solution, regardless of whether you buy the hardware in the DIY NAS: 2023 Edition. To see if your hardware is likely to work well with TrueNAS SCALE, check out the TrueNAS SCALE hardware compatibility guide. You can also check out the TrueNAS Community Forums and/or TrueNAS Discord for tips and examples of DIY builds and more on containerized apps (Nextcloud, Plex, Syncthing, and more!) with TrueNAS SCALE.
Component | Part Name | Qty | Cost | |
---|---|---|---|---|
Motherboard | Topton N5105 NAS Motherboard | specs | 1 | $199.99 |
CPU | Intel Celeron N5105 | specs | N/A | N/A |
Memory | Crucial 32GB DDR4 3200MHz CL22 (2x16GB Kit) | specs | 1 | $72.30 |
Case | JONSBO N1 | specs | 1 | $145.00 |
Power Supply | SilverStone Technology 450W SFX PSU (ST45SF-V3-USA) | specs | 1 | $89.12 |
Boot Drive | PNY CS1030 250GB M.2 NVMe | specs | 2 | $32.65 |
SATA Cable | Rosewill 3-Pack SATA III Right Angle Cable | N/A | 2 | $6.99 |
TOTAL: | $585.69 |
Most years, I include a time-lapse video of the assembly, but I wound up needing surgery to repair a partially torn rotator cuff. It was rather difficult to assemble the DIY NAS: 2023 Edition by itself–there was no way I was going to be able to set up my overhead camera rig!
That being said, assembling the DIY NAS: 2023 Edition was fairly straightforward. Building small form factor computers isn’t easy, but this year’s DIY NAS was easy enough that I managed to do it without aggravating my bad shoulder!
The only changes that I wound up needing to make in the BIOS were changes to the boot order. That’s a really simple change that’s usually straightforward. But the BIOS provided on the Topton N5015 NAS motherboard proved to be a bit more challenging than most other BIOS versions. It wound up taking a little bit of spelunking through the BIOS menus to find the boot order. It wasn’t difficult, but it wasn’t as easy as it should be.
The night before my shoulder surgery, I finished assembling the DIY NAS: 2023 Edition, booted it up for the first time, kicked off Memtest86+, and then promptly forgot all about it as I recovered from my surgery.
Nearly 76 hours later, it’d completed a total of 42 successful passes!
Normally, I’m inclined to do some CPU stress testing. But seeing as how the answer to “the Ultimate Question of Life, the Universe, and Everything” matched the number of successful Memtest86+ passes, I decided that the three days’ worth of running Memtest86+ was sufficient burn-in all by itself.
As I’m writing this blog, the uptime on the DIY NAS: 2023 Edition is over 60 days and would’ve been longer had I not been keeping it up to date with updates to TrueNAS SCALE since I started building it. The machine has been completely stable.
Each year, I like to do a few different benchmarks to validate that the DIY NAS is performing up to my expectations. I’m not particularly interested in tweaking for the absolute best performance; these benchmarks are more a matter of validating that what I built is sound.
The first thing that I benchmark is network performance. The first and most likely bottleneck for a DIY NAS is going to be the network. TrueNAS SCALE includes the iperf3 binaries, so I grabbed the Windows binaries for my desktop PC and directly connected the DIY NAS: 2023 Edition to the unused 2.5gbps interface on my desktop computer. After setting up static IP addresses on each end, I ran iperf3 both as a server and as a client on this year’s DIY NAS. Unsurprisingly, iperf3 was able to fully saturate the 2.5gbps connection between my desktop PC and the NAS.
After that, I skip right to benchmarking throughput to a drive mapped to a Samba share hosted by the DIY NAS: 2023 Edition using CrystalDiskMark. I was pleased to see that the 2.5gbps network connection was pretty much saturated on both sequential reads and writes to the drive I mapped to an SMB share on the DIY NAS: 2023 Edition.
In the past two years, I dove headfirst into home automation. One of the things that I’ve done in the past year is put smart outlets at each of my computers for monitoring their power consumption and track them using Home Assistant’s Energy Management. I made sure to reserve a smart outlet for my DIY NAS builds too.
For fun one day, I worked through a series of tasks and noted the peak wattage during each of those tasks. In addition, I captured the average wattage of the NAS being idle for an hour. Here’s a graph of the wattage over the 24-hour period from the beginning of these tasks.
I went through some of the metrics captured in Home Assistant and the peak wattage recorded during each of those tests. I also grabbed the peak and average wattage when the NAS is idle. Lastly, I grabbed some data for a 36-hour period of time–including these tasks–and captured the maximum wattage, average wattage, and total consumption over those 3 days.
Task | Duration | Max Wattage | Avg. Wattage | Total Consumption |
---|---|---|---|---|
Idle | 60 min. | 68.8 w | 59.87 w | --- |
Boot | ~2 min. | 73.0 w | --- | --- |
FIO Benchmark | ~13 min. | 73.0 w | --- | --- |
SMB Benchmark | ~4 min. | 77.6 w | --- | --- |
ZFS Scrub | ~1 min. | 71.2 w | --- | --- |
S.M.A.R.T. Long Test | N/A | 76.2 w | --- | --- |
Total | 77.6 w | 60.49 w |
Notes:
- The ZFS scrub was comically fast because the pool was empty. I don’t believe that what I’ve captured here is a quality measurement; please take it with a grain of salt!
- A commenter on the DIY NAS: 2023 Youtube video spotted an issue with this power consumption data. The average wattage of 60.49W and 3.4kWh of consumption in 36 hours conflict. An average of 60.49W over 36 hours should be ~2.2kWh or 3.4kWh over 36 hours should be an average of ~94.4W. I’m sure this is a result of error(s) on my part, but I don’t have the ability to go back to identify which error(s) were made and correct them.
Lastly, I wanted to share just how much less power the DIY NAS: 2023 Edition is using compared to my own desktop computer (Animal), my own TrueNAS SCALE machine and my homelab machine (Deskmeat). So I charted out the power consumption of the 24 hours following the beginning of these tasks.
It does not matter. Regardless of how much effort I put into parts or how carefully I work to assemble the DIY NAS, there’s always something that I believe can be improved with every DIY NAS that I build. The DIY NAS: 2023 Edition is no exception.
Of these complaints, the first one is the most worrisome. Upgrades aren’t actually impossible, but if you’re wanting more CPU, or PCI-e slots for other components, then you’ll need to shop for a new motherboard. The other two weaknesses seem a bit nitpicky.
The cost of the four prior DIY NAS builds without any storage was $1565 in 2022, $1728 in 2020, $1379 in 2019, and $1890 in 2017. At $630, the cost of the DIY NAS: 2023 Edition is between 33% and 46% of the cost of those systems!
Each year, I strive to put together something with a quality price-to-performance ratio, and I think the DIY NAS: 2023 Edition really hits that mark. I think it could easily be argued that this year’s build is one of the best in terms of its value.
It should go without saying that I’m impressed by the DIY NAS: 2023 Edition, but what do you all think? If you’re interested in building your own DIY NAS, does the DIY NAS: 2023 Edition seem like something you would want to build? Has the DIY NAS: 2023 Edition sparked your curiosity and inspired you to build a unique DIY NAS of your own? Let me know in the comments or come tell us about it in the #diynas-and-homelab channel in The Butter, What?! Discord server.
As I have more than 10 times in the past, I’m giving away the DIY NAS: 2023 Edition and a $100 giftcard for some TrueNAS schwag! The team at iXsystems deserves a shoutout for sponsoring the giveaway and spicing it up! In prior years, this was a luck-of-the-draw raffle, but I’m changing things up this year.
For this year’s giveaway, I’m tasking entrants with answering this question:
If you won the DIY NAS: 2023 Edition, what would you do with it? What sort of problems would it solve for you?
Share your answer somewhere publicly accessible on the Internet: in a blog, a Tweet, post it to Mastodon, talk about in on TikTok, vlog about it on YouTube, live stream it on Twitch, etc. Then submit your entry by filling out the DIY NAS: 2023 Edition giveaway form.
As the giveaway runs its course, I’ll be consuming each entry, promoting everyone’s entries on social media, and eventually picking a winner. The winning entry will find a way to set itself apart from the others. The creativity and quality of the entry will be critical! Please keep that in mind when choosing the platform and crafting your entry.
As an example, it’s hard to imagine that a Tweet that says “To store files!” could be more compelling than a YouTube video from a dog rescue group that needs more capacity for videos that they create to help abandoned dogs find their forever homes.
Come talk to me in the #giveaway channel of The Butter, What?! Discord server if you want to maintain a bit of anonymity or if you lack a suitable platform to share your entry. We’d love to host your entry on The Butter, What?! blog!
The EconoNAS has always intended to act as a template for other budget-friendly DIY NAS builders, which is more capable and has a more attractive price tag than the off-the-shelf NAS products.
At several points in 2022, I’ve said to myself “I’m going pick out parts for the DIY NAS: EconoNAS 2022!” I started shopping online, compared it to the previous EconoNAS, became disappointed, and decided to set the entire build aside to work on a later day.
A number of factors exist today that continue to make an EconoNAS difficult to come up with. Corporate greed, inflation, a lack of new product offerings from Intel/AMD, supply chain disruptions, and the lingering impacts of COVID-19 have all worked against the concept of building an economical NAS which is unique from the prior year.
Each and every time I worked on picking out parts, I realized I was basically just rebuilding the DIY NAS: EconoNAS 2020 over and over again.
The last EconoNAS was one of my favorite NAS builds, but it was not perfect! My biggest complaint about the DIY NAS: EconoNAS 2020 was its price tag. Since publishing it 18 months ago, many of the components’ prices have steadily climbed to the point where I feel it at the border of no longer being economical.
The DIY NAS: EconoNAS 2020 might have been expensive, but it was quite capable. It was hands down the most capable EconoNAS that I have ever come up with. Even though it is quite expensive, its capability makes it a really good value.
Ultimately, I decided to take last year’s EconoNAS and reimagine it by focusing on making it as economical as possible.
Disclaimer: Just like last year, I did not actually build this myself. I’ve done everything else in my power to crosscheck the compatibility of these parts. If I were building an additional NAS, I would not hesitate to buy all of these parts myself. Hopefully somebody can share their experience building something similar down in the comments below!
Since the Athlon 3000G’s initial release, AMD has yet to release another AM4 CPU which seems equally (or better) suited for an economical NAS build. I was a little disappointed that AMD’s latest CPU release did not include a comparable inexpensive dual core and four thread CPU with a reasonably low thermal design power (TDP).
So just like last year, this year’s EconoNAS is built around the AMD Athlon 3000G CPU. Among the Athlon 3000G’s specifications, these are the specifications that I think make it well suited for an economical NAS:
If I didn’t change the CPU, then why change the motherboard? The only reason I could think of for changing the motherboard would be if there was a significant price difference.
The product listing for the Gigabyte Aorus M (specs) has hovered right around $100–right near where its price sat at the end of 2020. As I worked on this blog, I saw the price go down as low as $79.99. Combined with its price, this motherboard’s specifications still make it ideal for the EconoNAS:
In the previous EconoNAS build, I chose to get 32GB of DDR4 RAM. This choice was quite extravagant for a budget-friendly build. The focus for this year’s EconoNAS is budget, so I decided to trim the amount of memory.
At first, I was tempted to go with 8GB of RAM because that’s more than enough to meet the minimum requirements of the TrueNAS products, OpenMediaVault, and UNRAID alike. However, more RAM in a NAS is beneficial enough that picking a 16GB kit (2x8GB) of Crucial DDR4 2666MHz RAM (specs) seemed like a good value.
When it comes to being economical, picking out a boot drive can be especially difficult. Especially when many NAS products use the entire boot drive’s capacity. Higher-capacity NVMe and 2.5” SATA SSDs are the best value (dollars per GB), but much of that potential value is wasted if the entire drive is consumed by the NAS distribution. There are many viable options; here are my two favorites:
The second option is a bit more complicated and often not recommended (or supported) by the NAS products. So for this year’s EconoNAS, I opted to go with the first option and selected a 16GB Intel Optane NVMe SSD (specs). At a little over $1/Gb it is not a tremendous value, but the capacity is sufficient for its purpose, it is relatively inexpensive, is high performance, it frees up a SATA port, and frees up a drive bay.
Update! Thanks to an eagle-eyed reader pointing it out, I’m realizing that I overlooked the fact that the Gigabyte Aorus M motherboard manual clearly states that using an NVMe drive will prevent you from using 2-3 of the SATA ports. An inexpensive SATA SSD like the Kingston 240GB A400 is probably a much better idea for the boot drive!
As far as value goes, it’s hard to beat the case used in last year’s EconoNAS, the Fractal Design Node 804. It is a fantastic case and at the time it was selling for a fantastic price. To save some money in this year’s EconoNAS, I opted to go with the Antec VSK3000 Elite (specs), which is almost half the price of the last EconoNAS’s case.
With an inexpensive 5.25” to 3.5” drive adapter you can fit up to five 3.5-inch hard drives, a Micro-ATX (or Mini-ITX) motherboard, and an ATX power supply in this case. It’s a fairly small-ish form factor case from a reputable brand which helps make it ideal for this budget-friendly DIY NAS build.
For the power supply, I searched around a bit for an inexpensive 80 Plus-certified power supply that seemed to be well reviewed and found the Enermax Cyberbron 500W (specs). 500W should be an ample amount of power to supply to the CPU, up to five 3.5” HDDs, and the rest of the components in the parts list.
Component | Part Name | Qty | Cost | |
---|---|---|---|---|
CPU | AMD Athlon 3000G | specs | 1 | $79.95 |
Motherboard | Gigabyte B450 Aorus M | specs | 1 | $102.63 |
Memory | Crucial 16GB Kit (2 x 8GB) DDR4-2666 UDIMM (CT2K8G4DFRA266) | specs | 1 | $57.99 |
Case | Antec VSK3000 Elite | specs | 1 | $70.60 |
Power Supply | Enermax Cyberbron 500 Watt 80+ Bronze | specs | 1 | $37.99 |
OS Drive | Intel Optane Memory M10 16 GB | specs | 1 | $18.25 |
TOTAL: | $367.41 |
The particulars of storage are really quite specific to the needs of the DIY NAS builder. As a result, I haven’t been including storage drives in my build suggestions. I do have a couple suggestions which I think will help you in choosing your storage media:
The answers to these questions should help point you toward the total capacity of your NAS, the level of hardware redundancy you’ll want to have within your NAS, and hopefully help you realize that you’ll likely need a way to potentially back up the contents of your NAS if you really care about your data.
All that being said, here are few decent deals on hard drives that could go well with the DIY NAS: EconoNAS 2022:
If you’re looking for HDD deals, you can join the #deals channel in the Butter, What?! Discord server. I and others regularly share good deals in this channel that we find elsewhere.
Hard Drive | Price | |
---|---|---|
HGST Ultrastar He10 (HUH721010ALE600) 10TB (Renewed) | specs | $77.50 $7.75/TB |
Western Digital Ultrastar DC HC520 (HUH721212ALE600 ) 12TB (Renewed) | specs | $118.99 $9.92/TB |
2-pack Seagate Exos X16 16TB ((ST16000NM001G) | specs | $433.99 $13.56/TB |
Another change I’d make in the 2022 EconoNAS is to change which TrueNAS product is in use. For this build, I would highly recommend TrueNAS SCALE. There’s a lot of reasons to like TrueNAS SCALE, but above all others I think that the fact that TrueNAS SCALE is built atop Debian makes it ideal for an Economical NAS build. The hardware support for consumer-grade hardware is much better in Debian that it would be under FreeBSD (used by TrueNAS CORE).
TrueNAS CORE and TrueNAS SCALE have been pretty comparable for my own use. However, I think it is worth pointing out that with regards to file sharing, others (and I) have found that TrueNAS CORE outperforms TrueNAS SCALE in benchmarks. With regards to actual real-world use, I have not noticed this performance at all on my own DIY NAS running TrueNAS SCALE.
All that being said, I fully expect that TrueNAS SCALE is more than capable of saturating the gigabit network interface found on the motherboard.
At its roots, the DIY NAS: EconoNAS 2022 started out as a carbon copy of the last EconoNAS. This time around, I put an emphasis on shaving dollars off the price tag. I’m a tiny bit disappointed that I ended up recycling old material and I’m also a bit disappointed that this EconoNAS can only accommodate five 3.5” hard drives.
Everything that I was excited about for the prior EconoNAS is still valid. Thanks to the AM4 socket and an up-to-date BIOS, there’s a ton of room for CPU upgrades. The two empty DIMM slots mean that RAM upgrades are possible too. The empty PCI-e slots mean that a 10Gb NIC, an HBA to support more HDDs, and potentially a GPU could be added to make this economical NAS even more powerful.
Most importantly, I’m excited that I managed to shave more than $275 (41%) off the price of the prior EconoNAS. My goal with every EconoNAS is to build something–storage included–that is $500 or less. You could buy three refurbished 4TB HGST enterprise drives with the $125 left over. A 4TB-12TB, 5-bay NAS, with lots of room for future upgrades for under $500 seems like a great deal to me.
I always worry that I might be a bit biased, so I spent a couple hours digging around the Internet looking at off-the-shelf NAS products. I searched for NAS machines with 5 drive bays and also for NAS products that are less than $400. Here’s what I found:
The 5-bay NAS products that I found were all considerably more expensive than this EconoNAS. But many of them also included premium features like 10GbE or 2.5GbE network interfaces. Considering how easy it is to build an inexpensive 10GbE network, I don’t think those features justify their hefty price premiums. The majority of these more expensive NAS machines also failed to measure up to the AMD Athlon 3000G and 16GB of DDR4 RAM in the EconoNAS. Comparing this EconoNAS build to the products that were a similar price ($300–400) wasn’t even a fair fight. All of these machines had four (or fewer) drive bays, meager CPUs, and a fraction of the RAM.
In my opinion, off-the-shelf NAS machines’ only two benefits not accounted for in the EconoNAS are product support and hot-swap drive bays. As infrequently as you’ll need to swap HDDs, I think this a NAS’s most overrated feature. I think there’s definitely value in having a product support team to contact if you have questions, but that value diminishes pretty quickly if you’re willing to type the same question you’d ask into Google and read through the results that come back.
But enough of what I think! How much would you expect to be charged for a NAS with the following specifications?
If you were building it yourself, what improvement would you want to make on the DIY NAS: EconoNAS 2022? A bigger case? A more powerful CPU? What about more RAM? The biggest advantage of a DIY solution is that you get to make these decisions on your own rather than let a company make that decision for you.
Let me know what you think in the comments below or come share your opinion in the #diynas-and-homelab channel in our Discord server!
]]>But the Aufero Laser 2 wasn’t the only thing included in the package for me to review! Next to the laser engraver was the Ortur YRR 2.0 Rotary Roller. The rotary tool was so interesting to me, I felt it needed its own blog!
A rotary tool is an accessory to be used with a laser engraver. The accessory is comprised of a pair of rollers. This rotary tool is plugged into the laser engraver’s motherboard in place of the Y-axis stepper motors.
As the laser engraver sends commands to move the Y-axis, the rotary tool’s wheels spin and rotate your material while the laser moves back and forth the X-axis and engraves the material.
It takes some configuration and tweaking in the laser engraving software (LightBurn or LaserGRBL), but once that setup is done, the rotary tool enables users to engrave cylindrical objects like mugs, glasses, tumblers, baseball bats, etc.
The Ortur YRR 2.0 Rotary Roller comes in a kit that comes completely disassembled. Assembling that kit is fairly straightforward thanks to the documentation, Ortur’s assembly video, and other various guides shared by content creators on the Internet.
I assembled the entire thing after breakfast one Saturday morning. The assembly was complete before I had finished my morning cup of coffee.
While assembling the Ortur YRR 2.0 Rotary Roller seems to be well-documented, hooking it up to your laser engraver, configuring your laser engraving software, and the positioning of the rotary tool aren’t part of what’s covered in any documentation from Ortur. This is my only complaint about the Ortur YRR 2.0.
Considering that both the laser engraver and the rotary tool are made by the same company, I expected that there would be a myriad of directions for using the Ortur YRR 2.0 with all of Ortur’s different laser engraving products. I was a tiny bit disappointed when I realized this, but in retrospect I’m a bit glad these directions weren’t included! I enjoyed tinkering with the engraver and gradually coming to an understanding of how it works.
Thankfully, there’s a bunch of content from other makers that have put in the work to help fill this documentation gap. In particular, The Louisiana Hobby Guy shared a guide setting up the Ortur YRR 2.0 which I found tremendously handy.
I enjoy when my interests overlap. I especially enjoy when they happen as blog topics. For example, 3D-printing parts for my quadcopters, using home automation to monitor my 3D printing, and most recently 3D-printing parts to use with my laser engraver as part of this blog!
I wound up encountering two challenges with the Ortur YRR 2.0 Rotary Roller that I knew right away I wanted to solve with 3D printing:
I was able to conquer both of these challenges with my 3D printer; I even designed one of the solutions myself.
Firstly, I used these awesome Lego-like risers by Enduro512 on Printables.com. There were a variety of heights for the risers which make adjusting for different diameters of material a breeze. Secondly, I wound up 3D-designing a base that the Ortur YRR 2.0’s roller guide can attach to. This way the roller guide could be placed independently of the Ortur YRR 2.0 and support a much longer piece of material.
Here's some behind-the-scenes content for my next #laser #engraving content. I put my #3Ddesign skills and #3dprinter to use to keep this baseball bat level before I try and engrave it.
— Brian Moses (@briancmoses) August 29, 2022
Do you think it'll work? pic.twitter.com/GBMA6sj4q5
Want to buy the Ortur YRR 2.0 Rotary Roller but you’re a little wary by the lack of documentation on configuring your software and using the YRR 2.0 for the first time? Don’t be. There’s information out there that fills this gap, and if you’re willing to tinker, then it’s not that difficult to get set up without that documentation.
I tinkered and muddled my way through getting the Ortur YRR 2.0 working and thought I’d share a few tips that I think might be helpful for you too:
Complement these tips with others’ content, experiment with the Ortur YRR 2.0, and you will be well on your way to understanding how to get it to work best for you!
While I would’ve liked if there had been a bit more documentation to help me get started, I also enjoyed the extra tinkering with the Ortur YRR 2.0 that I needed to do in order to get it working. I wound up investing some of my raw material and time in the process, but I’m pretty happy with the returns from that investment.
Having an Ortur YRR 2.0 enables you to engrave an entirely different shape of material. It accomplishes this fairly inexpensively, without a tremendous amount of effort to assemble or effort dialing it in. I do not have a tremendous amount of experience with laser engraving, but I had it working fairly quickly.
If you own a laser engraver and you think it’d be fun to engrave mugs, drinking glasses, baseball bats, tumblers, stemware, etc., then I would encourage you to strongly consider the Ortur YRR 2.0 Rotary Roller!
]]>