ConnerVT

Guess that's okay if you have a place to secure it, out of the way of it impeding airflow. And I would want that to be secure, as the metal bracket falling on electronics would not be ideal. I find the cable with rubberized cover around the Type-A connector more versatile. A zip tie or two can secure it where you want, out of the way. If your server behaves, you shouldn't need to unplug it for years.

I ran a 1st gen 1500x in my server for over a year. The only two things I needed to get it to run reliably was to set the speed per the charts in the FAQ (which are right from AMD documentation) to 2133 and set to Typical Idle Current. The later was definitely needed, with my failure mode being that the system would lock up (stall) before changing to that BIOS setting. My 1500X was first run silicon. I got it one day before the official launch day. (I was pumped to get it, as I was one of many co-workers who made that chip.) The Internet is forever. So if someone posts something, it gets quoted again and again, even when it is no longer valid. When the original launch of Ryzen in 2017 happened, AMD was very focused on the Windows OS, and the Linux community had a lot of catch up to do. There were issues with Ryzen and Linux then, but both AMD and Linux have caught up considerably. the EPYC processor now sells as well as Intel processors in the data center space. This would not happen if the AMD processor was still flawed. Ryzen's Infinity Fabric (which handles all memory transfers) is very tightly tied to memory hardware, and is a bit less forgiving than Intel's memory handling. So you can't push it to the marketing department's claimed memory speeds. When they say "DDR4-3200" the 3200 is the XMP speed, what they tested the RAM to run. But that's the rating of the RAM, not the memory controller running it. You can put 150MPH tires on a Prius, but you'll never get it to go that fast. My experience has been that Windows is a bit bore forgiving than Linux when it comes to memory errors. (Not sure that's a plus for Windows or not.) Running my server at JEDEC speeds vs XMP may decrease performance, but that is more than offset by stability.

Or this cable, which I use in both of my servers. I prefer the cable to the adapter, as it allow you to locate the flash in a spot out of the way and not likely to get bumped if you need to go into the case.

For the Unraid build, I started with some of the components coming from my desktop PC upgrade: Ryzen 1500X, mATX MB, my first NVMe drive, and 8GB of DDR4 I had bought for another PC (now decommissioned). For ~$600, I bought a case, 4 re-certified 6TB drives, PSU, 8GB more DDR4, and some cabling. The case, power supply and cables may be all that is still in the server. I've made a lot of changes in the 2 years I've been running Unraid. That 1500X/MB/DRAM combo, and 3 of those 4 drives ended up in a backup server I just built this week. It all ended up in the case I had replaced when I updated when I updated the desktop PC I mentioned at the start of this post (just realized that as I typed this). More confusing than a hillbilly family gathering. 🤪

Last reply in this thread from someone associated with Limetech was April 3.

The build has begun. This may be the least expensive system I've assembled (ignoring the fact I did already spend $$$ when I first acquired the parts). Nearly everything is from upgrades done to other systems around the house. I bought a mid-grade power supply, an 80mm fan and a header to USB cable. Having 16TB of drives and not having a use for them is not usually a problem I've had in the past. Building in a Silverstone SG11 Sugo case. It was originally my daily driver's case, but the cooling isn't that great without really spinning up the fans. Really goes against the main reason to build a SFF/MFF system. It sits next to the desk in the family room, so it needs to be quiet. Fan noise is not an issue on the server bench in the basement. The build went smoothly, except for one SATA cable I forgot to attach to a drive, and the usual "Why can't I get this *#@% flash drive to boot??!?". All now resolved. Done messing with it for today and I will pick back up on it tomorrow. The operation will move downstairs, to put the system on the network, get basic Unraid functions working and start configuring it for its backup functions.

For me, this is kind of a funny story. It is difficult to give an accurate cost, for my server should be renamed Theseus. I originally decided to build an Unraid server with the $600 COVID stimulus check I received two years ago. I had a retired CPU, motherboard, power supply and some DRAM on hand, so I figured I would give Unraid a try. I initially came pretty close. Then decided I didn't trust the power supply, so I replaced it. Then bought two more drives. I really started to enjoy the server, and started to put it to work more. Bought it a GPU. And a UPS. So I ended up upgrading a few things. I upgraded the: NVMe drive used as the main cache pool/appdata CPU Motherboard (old MB didn't play well with the new CPU) Additional DRAM 4 more, much larger drives Right now, about the only things left from the initial build are the case, a 3x4 drive cage, the small ssd I have that will never die, the flash drive, some fans and cables. So, do I add up the cost of the original component, of the current components, or of everything that has passed through this setup? 🤔 One benefit of all of this is that I'll be starting a new Unraid server build this week. Other than needing a power supply, all of the remaining components I have on hand, many of them previously installed in the server above. So the total hardware cost will be about $70 USD. Of course, that's before upgrades. 😝

All solutions require they pass the WAF test.

No worries. It is a broad subject, so it isn't hijacking as long as you are bringing something to the table. You offered food for thought to me with your S3/S4 sleep comment. Something to look deeper once I reach this point. As far as to "what" I back up, first need to say what's on my server. It has backups of my other PCs and devices, about 20TB of media, an an aggregate compilation of my family's photos. I also have a NextCloud instance, but am not very impressed with its performance, and the data it holds is duplicated elsewhere, and I will likely just blow it all away anyway. Appdata, libvirt and flash are backed up weekly (old Appdata backup plugin, I'm still on 6.10.3) Photos and music media are on an external USB drive, kept off site. Very static, I tend to bring the drive home and update it every few months. (rsync) Backup files, as well as Appdata, libvirt and flash, are backed up weekly to an external USB drive (rsync). Each month, I swap that drive with another kept offsite. The gap I have in my backup of my media (movies and TV). It is all replaceable, but would be a heartbreaking loss. I have 18TB of drives now sitting in a drawer, all too small to be of use in my current server. It will hold my TV and (1080p) movies with plenty of room to spare in the backup server I'm planning.

In my opinion, it is not a SanDisk issue. Nor Samsung. Nor Kingston. It is a counterfeit issue. So it isn't the brand, it is the distribution network that can't be trusted. Amazon, eBay, Alibaba can't be trusted. Go to an actual brick and mortar store. Staples. Target. Walgreens even. Pay an extra $5, and save yourself a lot of grief.

Old post, and I'll reply in the wrong language. I have this SG11 case. Was my daily PC, but I switched to a Fractal Design CORE 500 as the Silverstone just has a poor airflow design. For my desk PC, I wished for it to be quiet, which was difficult to maintain low fan speeds and keep the system cool. I am planning to use this SG11 to build an Unraid backup server, with 3 6TB drives. Noise isn't an issue, so it will have the 120mm and 2 80mm fans, all running high speed. I hope cooling won't be an issue this time.

Neither count towards the number of drives for your license.

Excellent! Exactly the kind of conversation I was hoping this thread would produce. Ipmi would make for a neater solution, but I'm not buying any more hardware for awhile. Just dropped $$$ on 4 16TB drives (that's why I have the drives for this build). I did pick up a PiKVM from someone on this forum (*wink*), but it is working wonderfully fulfilling its planned purpose on the main server. Let's see who else may chime in with some backup server solutions.

I currently do the same for what are effectively the backup of my other computer's backups. PCs/devices back up onto my (main) Unraid server, then those backups go onto an external USB drive weekly. Every month or so, I swap the USB drive with another, which I keep offsite. Belt, suspenders, and some duct tape. I use a script for this, similar to Hoopster's but much more rudimentary. I plan to borrow some from what he posted, thanks for that. The backup server I plan to put together will hold a much more static (and less un-replaceable) data, so once a month sync with the main server. Curious how folks handle things without an ipmi to do wake/sleep/login on the destination server.

Thanks for the input so far - keep it coming! I definitely am looking at the backup server to be a part time operation. It will be built from retired parts. A Silverstone SG11 case (Sugo) and 3 non-He HGST 6TB drives I replaced in my current server. The SG11 was the case for my daily driver PC, retired as it has a poor cooling design (so not great for a system sitting next to my desk) . The non-He drives run about 3 to 4C warmer than the He filled drives that replaced them. Running the server, even for a day a week with the fans all wound up is OK by me. Running 24/7/365 just isn't desired, especially as the data it will hold isn't non-replaceable and I have no backup of at this time. Almost all of the data is better than none in this situation. I know there are folks here on the forum that have automated their BU solutions (be it by WOL or ???). Would love to have them chime in with how they sync things. I imagine there are a number of issues, such as determining the remote server is ready, when backup is completed, and powering down.

Well, not so weird if you say "40" afterwards. 😀

You would probably be better off posting in the [Plugin] Nvidia-Driver support thread. The GTX780 is a pretty old card, and not supported by the current Production Branch of the Nvidia driver. You would likely need use the v470 legacy driver.

You did great, Andrew. Thank you so much for sharing with us.

^ Agree. I played with the idea of putting opnsense on my server. Great idea in theory, but the devil is in the details. If the server is shut down, or even the array stopped, your VMs also stop. So everything works great *when* everything is working great. When something goes wrong? If your router is down, you can't access it, as your network is down. I know a lot of people virtualize their pfsense/opnsense firewalls. But nothing beats it running on bare metal. Makes troubleshooting so much easier, especially as you usually are without useful resources (such as your normally used network) when you find yourself trying to fix things.

I've just about completed the upgrades to my Unraid server. Was taking an inventory of all of the spare hardware I now have, and it got me thinking. Perhaps it is time to build a backup server. I already have a solid backup strategy for all of my data, except for my media. This backup server would be just the place to plug that hole. My hope for this thread is for the folks here who have an Unraid backup server to share their solutions of backing up one server to another. For me, ideally it would back up my TV and (1080) Movie folders which are part of my Media share perhaps once a month. The backup server would be powered down until it is time to sync the two. Saves power and helps isolate the backup from any bad actors. If you have a setup like this, please tell how you implement it. Do you run a script or utilize a docker? Run both servers on your LAN or direct connect the two? Does your backup server run 24/7 or do you power it up/down as needed, and if the later, how do you accomplish this? I am really interested in hearing how you approached this, as I'm just at the beginning steps of implementing something here.

By "Turning off" do you mean it powers off, or is it powered on and frozen/locked up?

Isn't that what I linked? 😝

Your memory setting is wrong, and the most likely reason for your crashing. You cannot use the XMP settings, it is running your memory much too fast. Go to the FAQ entry listed below. The 2200G is a first generation Ryzen. Your Corsair Vengeance LPX DRAM is almost certainly dual rank (You never know with Corsair...). So you would set the speed in your BIOS to 2400 (or even 2133 for troubleshooting). Running a MEMTEST wouldn't be a bad idea either. Don't believe the "3200MHz" the marketing people put on the package. You can put racing tires on a Prius, but it still won't go 120MPH.

No, it happens here with 6.10.3 as well. It is a bit disconcerting that this has been an issue that has prevailed for over a month and not been resolved. Running unraid-api restart is not a solution, but is just a Band-Aid that resolves the immediate symptom. My Servers has been sold to the Unraid community as one of the preferred secure methods of accessing a server remotely. By using My Servers you are handing you trust to someone else that they are keeping you secure. If this visible issue has been going on for a month, what else may be an unknown problem?

A quick follow up... Moved data off a 6TB drive, removed that 6TB and replaced the 8TB parity drive with a 16TB. Used that 8TB drive to replace a 6TB. Replaced 2 more 6TB drives with a pair of 16TB drives. Over 100 hours of parity building and data drive reconstruction. Zero errors. I really do like this server.