Miss_Sissy

There are people who strongly doubt that we actually landed a man on the moon, but it doesn't make it any less true. 😀 When you return a failed $7 thumb drive to Amazon, Best Buy, Target, etc., it does not get flown back to China, South Korea, etc. for failure analysis and teardown. It either gets tossed in the trash or sold on a pallet full of untested customer returns. No manufacturer sees $7 flash drives again after they leave the factory, whether they are genuine or counterfeit. Limetech said that they don't know if the non-serialized (via GUID) SanDisk flash drives are genuine or counterfeit, but SanDisk will not respond to them. Do you have information indicating that genuine SanDisk thumb drives all still have unique GUIDs? https://exceptionaltiming.com/15-ways-to-know-if-a-timex-watch-is-an-original/ But I agree with that general point you were making about the price. I could understand counterfeiting high capacity thumb drives, where the retail price is high enough to make it worthwhile. But Best Buy has a 32GB SanDisk - Cruzer Glide thumb drive for $7.19. I'm just not seeing any real margin at that end of the market, especially since the counterfeiters would have to amortize the cost of molds over such a small number of drives (relative to what SanDisk sells). If you are a Chinese counterfeiter, you're much better off remarking $0.50 NE5532 opamps as something like $6 OPA2134s. No retail packaging required. No plastic molds. Hundreds of them can be shipped in a few IC tubes. And there is almost zero chance they will be spotted as counterfeit at the port of entry.

Even some flash drive models that had unique GUIDs in the past now do not, likely due to late-stage capitalism squeezing manufacturers to save fractions of pennies per unit. (Should we start a thread listing models of computer cases that do not have removable drive caddies?)

I must have missed that part of your earlier message. Didn't you write that USB flash drive failure is a foregone conclusion ("When (not if) they fail") under Unraid? Thanks for sharing your thoughts, but I have locally hosted public and private network services for over 23 years.

Competing 4GB industrial, SLC drives from Amtron, Delkin, and Innodisk range in price from $58 to $86. If those were 'insane prices,' Avnet, DigiKey, and Mouser would not carry such products since those distributors cater to engineers who know what products like those should cost. It's a SLC-based SSD with 10-year data retention, 60,000 program/erase cycle endurance, and an MTBF of >5,000,000 hours. It has a real SSD controller with wear leveling. It's a completely different class of device than a mass-market thumb drive. Then why do so many people with name-brand, consumer-grade flash drives report failures when using them as Unraid boot drives? Unraid is not an application that writes much to the drives. It works fine if you're sitting at home with time to kill when you notice the failure. But what do you do if you are on travel and get a panicked call from your spouse saying they can't access files on the NAS they need for their work? What if you're running a public-facing mail server, web server, or FTP server? I'm not trying to convince you to move to a high-reliability/longevity industrial SLC drive if you don't feel that you need it. But the price of the recommended drive is not at all out of line for that class and capacity of storage device. ☮️

That’s exactly my point: Samsung has the flash tech to make long-lived USB drives, but they only offer it in their PRO Endurance line of MicroSDXC cards. That’s probably because most USB flash drive buyers care only about speed and price, not longevity.

In the U.S. they are not very expensive, with the 4GB version selling for about $55 (U.S. dollars) on DigiKey.

I never suggested that there just be a one-time check at boot time. Unraid could periodically, or randomly, read the GUID while it was in operation. In fact, I proposed using a USB flash drive that might mot even be formatted and writable — it would just be the source of a GUID. Absolutely unacceptable.

I've been using a Toshiba 16GB USB 2.0 flash drive on my current Unraid server. It self-identifies as "TransMemory" and the drive's pronoun is "it." If the drive becomes non-binary, I have a real problem, but at least no one has to worry about which restroom it uses.

Good advice, but I'm looking for something reliable when I'm on an extended trip far from home. Then if I'm hosting publicly facing services on my Unraid server, such as VPNs, web servers, or FTP/SFTP servers, they won't suddenly disappear for days or weeks until I return home and resuscitate my Unraid server. I suggested allowing Unraid customers to boot from any drive they wanted as long as the licensed thumb drive was plugged in. People who are fine with the way things are now could continue booting from their thumb drives while others could elect to use more reliable media like SSDs, allowing the thumb drive to be used as a form of dongle to which no data is ever written and the only thing read is the GUID.

The main issue for most people was anecdotal evidence of USB 3.0 flash drives exhibiting a higher failure rate than USB 2.0 flash drives. Or maybe Limetech had statistics that I was not aware of. I want a flash drive equivalent to the Samsung PRO Endurance MicroSDXC cards. Those sacrifice a little bit of write speed to get much greater lifespans for write-intensive uses like dash cams, body cams, and security cameras. I use them almost exclusively as the bootable primary drives on Raspberry Pi web servers that I run. I have yet to see a failure.

I was a bit thin-skinned after a miserable day of home repair, so frustration got us both. Peace, friend! ✌️

No, it has not “gotten off in the weeds.” This was never a thread of brand-agnostic USB flash drive recommendations. It’s a warning about Sandisk USB flash drives and a follow-on discussion of alternatives, including SSDs.

I agree: Unraid needs an rsync GUI, especially to enable and configure rsync as a service. rsync has been a basic part of every NAS OS I've ever run or considered, including QNAP, TrueNAS, XigmaNAS, OpenMediaVault, Synology, and Thecus. On those, it wasn't something where you had to 'roll your own' by manually enabling and configuring the rsync service from a command line or choose between multiple Docker rsync implementations. This should be a pretty easy feature to implement, so I hope that Limetech sees fit to make it a standard part of Unraid.

Thank you for your courteous replies. Two points: Redundant RAID is not just for uptime. It's also the data protection for all of the additions, deletions, and modifications to your data since your most recent backup, all of which you would lose if you had to go into disaster recovery mode and restore everything from the backup. The 3-2-1 strategy (mentioned by another user) is nearly 20 years old and slowly falling out of favor as more robust, modern strategies with names like 3-2-1-1-0 and 4-3-2 gain prominence. See this Backblaze white paper and this Stonefly article which go into more detail than would be appropriate in a thread response here. Forms of RAID that provide snapshot capabilities, such as ZFS RAID, give you a recovery option without resorting to a separate backup. With a RAID mode that can survive individual drive failures, the snapshot capability relegates backups to something only pulled out when something catastrophic has happened (fire, flood, ransomware, etc.) or when you need to recover old (rather than recent) versions of files. Unraid and Stonefly are on opposite ends of the spectrum, with Unraid being software aimed primarily at DIY home networking enthusiasts and Stonefly marketing their products as complete "Data Center Solutions." Stonefly has been around for a long time and is well-respected. You should talk to one of their reps and get a handle on the true costs, not only the cost of the RAID appliance(s) you would host locally. Find out the cost of all ongoing services you would need from them, including cloud and support. Determine what your firm's role would be in installing, configuring, and managing the Stonefly hardware and software. Then think about whether you have someone on staff who can handle it and has the hours needed. Not knowing more about your firm, your data, and your budget, that's about all I can suggest. This March 13, 2024 review/article about Stonefly's offerings might be a good place to start. I do not know the author or whether he has any conflicts of interest. Good luck.

I'm glad that the design worked and I'm sorry that the parts don't provide a unique GUID for our use with Unraid. I was nervous that you couldn't do any iterations to make adjustments, but it sounds like the service you used has printers that 'agree' with mine.

As we have been asked by a moderator to cease debating in this thread, I will abide by his direction.

You are incorrect. Without redundant RAID, a single drive failure can lead to the loss of all data added or changed since the last full, incremental, or differential backup. Nice strawman. I never suggested that only catastrophes cause data loss. When I referred to "catastrophic data loss," I was talking about to the magnitude of the loss, not the cause. But I think that you knew that. Deleting a file or a directory tree is not a catastrophic data loss on any competently run NAS and network. It's an inconvenience that needs to be resolved through the use of snapshots, revision control systems, and other tools regularly employed on modern networks. Had you read the entirety of my post, you would have seen that I addressed the importance of backups, including offsite storage. What I did not do was propose a one-size-fits-all backup strategy that I claimed was "the standard" (without citations) for critical data. I'd need a lot more information from the client before devising a backup strategy to fit their business needs.

I'm going to give you some general advice because it sounds like that's what you are asking for. If you're feeling overwhelmed, I recommend that you hire a consulting engineer near you who has done this sort of thing before. When dealing with business-critical data, you don't want to experiment and try to learn on your own. Protection against drive failures is something best accomplished through RAID, not backups. For example, on our primary NAS, up to two of the six discs can fail without the NAS going offline or losing any data. On our secondary NAS, which has less valuable data, one of the five drives can fail without data loss. There are various forms of RAID and they can be configured to survive multiple drive failures. To use an extreme example, you could buy four 18TB drives and configure RAID so that you would lose no data even if only one of the three remained functional. Backups are for dealing with a catastrophic data loss, whether through flood, fire, malware, hackers, or some other act of God, Satan, or the Flying Spaghetti Monster. For that reason, you should follow some sort of backup strategy that puts backups offsite -- possibly in cloud storage or possibly on physical media. Based on what you've shared, my first thoughts would be an Unraid NAS that supports SAMBA protocol and automatic snapshots (snapshots let you roll back disks and directories to their state at some prior time). Unraid with a single ZFS pool consisting of four physical disks configured as RAID-Z2 would give you the abiliy to have half of the four discs fail simultaneously without any data loss, while supporting the aforementioned snapshots. My reason for suggesting Unraid is not because this is an Unraid forum. I'm on the Unraid forum because I've tried multiple commercial and open source NAS solutions and I think that Unraid is the best NAS OS, especially from a user interface perspective. I hope that I have left you with fewer questions rather than more. But remember that free advice is often worth what you pay for it, so don't trust me and especially don't trust anyone who disagrees with me!

I've been spooked off of getting 'creative.' The problem is that the first person who tries something will probably see success. But what happens if another person tries the same thing and it turns out that their device provides the same GUID numbers?

This is how we find things out. After the effort I put into designing the enclosure yesterday, I'll hang onto everything for general usage. Also, I don't want to get on Amazon's bad side for returning items that work as advertised.

Sorry about that. But it could be that it depends on revision or lot number of the adapter or the mounted mSATA SSD. You may be lucky even if I was not.

Warning: I bought both products from those links and got this GUID-used-or-blacklisted error when I tried to apply my activation code: So much for "appears to provide unique GUIDs for Unraid." I emailed Limetech support and am awaiting their response. I suggest that people hold off on buying these pending some sort of resolution.

After seeing your request to JonathanM, I created my own design which I am happy to share. It does not have any markings or trademarks, but it does have around 350 diamond-shaped ventilation holes for cooling. I've uploaded the .stl files to Thingiverse along with some images, notes, and settings. You can download the files and read more design here: https://www.thingiverse.com/thing:6520361 I hope that folks find it useful.

It takes more than just a few, inconsequential moments for me to get down on the floor and reach in to blindly detach cables and extract the NAS from the bottom shelf of a tightly packed unit in order to remove the cover -- and then reverse the steps to return the NAS to service. There's a lot more chance for accidental damage in that operation than there is in just swapping an external flash drive. No one in my home is going to reach behind the NAS to remove, steal, or damage the USB flash drive plugged into the back of it. If some stranger has broken into my home, I have much bigger concerns than whether they will take the flash drive out of my Unraid server. If installing the flash drive inside of the cabinet adds twenty minutes of extra downtime every ten years, then downtime was not minimized. I don't need more stress and delays when I'm already dealing with a server that's gone offline. Nor do I want to be the only person who can swap it since I might be on travel when it fails, just as I was when a NAS drive failed and I had to talk a family member through swapping the drive (illustrating the value of drive trays/sleds and spare, pre-cleared drives ready for installation). What you're suggesting has its merits and it's something I considered when I put the NAS together. But it was not ideal for me.

That's a bad idea if you want to minimize downtime. I have removable drive trays because I don't want to pull my NAS out of service and disassemble it when something fails. It's a lot easier to reach behind and swap a thumb drive in the blind than it is to demate all of the cables, pull it out of the shelf, put it on a workbench, remove the cover, swap the thumb drive, replace the cover, put it back into the shelf, and then blindly try to plug the power and network cables into the correct locations.