TimTheSettler

There was a time when I had to change this setting and I too had a hard time finding it. Settings like this should be shown but disabled, similar to other read-only text on this Docker settings page. Even better is an icon next to the value with a tool tip telling you that the value can only be changed when the Docker feature is turned off.

My post may or may not provide the exact answers you're looking for but hopefully it will provide some insight. I use Vorta which is a GUI for Borg. I chose this because I have a Linux friend who really liked Borg and he used it at work to back up their sensitive data. I use Vorta instead of Borg directly because I no longer use console apps, a well-built GUI is easier to understand. That being said, Vorta is a bit kudgy but it allows you to specify multiple source locations and a target location that can be local or remote. It uses deduplication and compression which both cut down on the size of the backup. It also uses encryption for the target location and it supports pruning which allows it to keep a certain number of older snapshots. However, keep in mind that deleted files remain until they are pruned. That means that if your pruning is set for 10 years then a deleted file will stay around for 10 years. I'm using an older Vorta docker template but the new one has a RESTORE option which allows you to mount an older snapshot so that you can restore files from that. Each backup/snapshot only keeps a note of what changed from the previous snapshot. Vorta runs each day for me (daily backup) so if nothing changes then nothing is added to the archive. If a file moves around within the source directory of the archive (moves around within the archive) then it's old and new locations are noted but no additional space is taken up. If a file is deleted it stays until it's pruned. I have different docker containers/archives for different types of files (documents, pictures, music, videos, etc.) because smaller, bite-size containers are easier to manage than one massive container (I have 20TB of data being backed up in 7 containers/archives). Note that moving files around in such a way that they move from one container/archive to another defeats the purpose of the deduplication since the file moving from one container is deleted from there (and will remain there until pruned) and that file will now appear in the new container thereby taking up space in both containers. Let me know if you have a specific question.

That checkbox is related to the "Check" button to the left of it. When you "Check" the parity you can let it check only (checkbox off) or you can have it write any corrections it notices while checking (checkbox on). If you run a manual "Check" (by clicking the button) you might want to run it without making any corrections first (checkbox off). If the check comes back and tells you that corrections should be made then you can decide to run it again with the checkbox on. As a programmer, I'm used to having a "test" or "check" mode that allows me to run the program without committing any data/changes. I see the results and if I like what I see then I run it again in "commit" mode.

Your hardware and future plans all sound good. No need to go crazy. Gradually add what you can. That's the beauty of unRAID. Everyone has a different idea about software but maybe the only "must need" thing is the "Appdata Backup". This backs up your docker containers. Also good to use Duplicati or Vorta to back up your data to another location.

"dispatching hogs" ... I'm imagining this...

Thanks. I'm going to try it on my test server. It seems to be missing some helpful values though.

@binhex Hi binhex. We need your help with syncthing v2. I'm not sure how a new docker will work if you need to run an upgrade process. I guess the script would need to be changed to do that if the current version is lower than version 2.

That's very cool. You should check out this forum post in the "Docker Containers" section and see about becoming a contributor. https://forums.unraid.net/topic/40696-why-cant-i-post-new-topics-in-here/

All files on all devices are synchonized to my file servers and those are syncrhonized with each other (using Syncthing). The backup server then creates snapshots each night of all that synchronized data (using Vorta/Borg). All servers are in separate locations. All servers use dual parity and cache nvme in mirror mode like @Joloxx9 .

You should plan out what you intend to do with your unRAID server before you get in too deep. What I mean is, how do you intend for it to be used? Will it be a file server? Will people have access to shared folders? Will it have apps running? Will those apps create, download, or manage data? Here's the setup that I'm running. If you have some app that downloads content then maybe it makes sense to create a share for that and then assign certain people to manage those files or maybe the data is dumped into a common folder that people have access to.

It looks like that mobo supports 4 NVMe cards which is nice but you only have 2 NVMe in your parts list. Maybe upgrade for later? The CPU and board look good. Intel LAN (2.5Gbps). Supports virtualization. Has three PCIe x16 slots. Supports 128GB RAM but has only 16GB. Not that this bad, it's ok. I think that 32GB would be nicer but you can always upgrade if you see the need. My servers all have 64GB and that's overkill. I have Jellyfin, Plex, Syncthing, and a few others including a VM running and my memory load is about 9GB. All in all, it looks good.

Why do you have a blue "X" over the two front fans? Does that mean that there are no fans there right now or that you can't put any fans there at all? I always try to have the same number of fans pulling air into the case as I have pushing air out. That being said, you should have four fans in the front pulling air in and then have the two back fans pushing air out and two fans on the top (on the mobo side) pushing air out of the top. Then be sure to have clearance in the front, the back, and the top so that the fans can pull the air in and push it out ok. Lastly, being in a closet doesn't help because the air in the closet will heat up and can't be refreshed with new, cooler air. It might be best to get it out of the closet. So, first thing is to add the fans as I mentioned and give the case some clearance to pull air in and out. See if that does it. If that helps but everything is still hot then move the case out of the closet.

Looks pretty good. Will you be using any parity drives and if so, how many?

It seems that the app was removed from the app store but the docker container is still maintained at the link below. I believe that someone just needs to add it back to the app store but I'm not sure how to do that. https://github.com/borgbase/vorta-docker

One other thing to add... If you're writing to the array while the parity check is running then everything will slow down. Here's a bit of a comparison for you. Notice that my last check was quite slow because a massive sync happened to be running when the scheduled parity check kicked in.

The point of my post was not to complain or panic about the reallocated sectors. I've run into this problem many times in the past and I usually acknowledge the error and leave the drive as is. I'll replace the drive only when the event shows up again. That's why I like unRAID. I have my parity disks as insurance and I can monitor the situation. I know what you mean but if you have two parity disks then identifying and fixing a problem is possible. Let's say that the parity has discovered an anomaly. If there are two parity disks then there are three sources of information. If two of the three match then it's a safe assumption that the majority is right. The third disk should be changed to match the other two, whatever disk that might be (the original or one of the parity disks). I like the idea of having parity to replace a disk but it should also be used to verify a disk if/when that disk has a SMART error. This would allow you to verify if the disk is still ok or at least that the data is still ok until the next event. The person then has some time to replace the disk.

Is there a way to verify and correct a disk from parity? I know that the parity check verifies the parity disks but I have a disk that has reallocated sectors. I have no intention of replacing the disk at this time because this could just an anomaly but I'd like to verify that the disk contents are ok. So, to me, that means that I should be able to use parity to verify the disk. Is there a way to do this?

You can create your own docker or ask one of the people who regularly create dockers to create one for you.

There's really no point to this in a home server scenario. It's expensive and you need a mobo that supports it. To me it's overkill and not worth it.

There are two ways to do this. I've done both scenarios, especially #2 where I was running Windows, set up that system to run unRAID, then briefly switched back to Windows. Scenario 1 Build a completely new machine. Make sure that the two machines can talk to each other. Copy your data from the old machine to the new machine. This gives you a backup system in case something goes wrong or if you want to start over. Scenario 2 Remove the Windows C drive and set that HDD aside or leave it in the case but leave it disconnected. Add your new memory and drives. Start the machine using the unRAID boot disk. Configure the system as you like it. Connect the Windows C drive to unRAID as an unassigned disk. Copy your data from that C drive to your new unRAID server. If worst comes to worst you can disconnect all the drives (including the unRAID boot drive) and reconnect your C drive. In this case the Windows drive will boot into Windows. You can then disconnect the Windows C drive and reconnect all your other drives (including unRAID) and then you'll have unRAID back again. It's important for Windows to only see the Windows C drive.

You only lose the SATA port if the NVMe is a SATA-based device. Read this --> https://www.kingston.com/en/blog/pc-performance/two-types-m2-vs-ssd and this --> https://www.atpinc.com/blog/what-is-m.2-M-B-BM-key-socket-3

It's stuff like this that's the reason why I love unRAID.

I think this is where you're going to see an improvement. My file servers (i3-10100) are slightly better than your existing CPU but my dockers are sitting in a couple fast NVMe drives. At least for me I don't see a problem. My plex works nicely. That's a good question. I don't have room in my servers for a drive to just sit there so I have a couple sitting on a shelf. To me that avoids unnecessary heat and power usage but sitting on a shelf usually isn't good over the long term. At the very least it's smart to spin the drive up every 6 months or so. That helps to keep the components in working order. I just connect it to an external drive bay, let it spin for a little while, then remove it and put it back on the shelf.

Any computer will work with unRAID. Find an old computer or build a new one. It's up to you. My test server is an old Windows desktop that I repurposed. The beauty of it is that you don't need the same size drives, mix and match as you see fit. Right now I see that the 14TB drives are the best price. Grab a few of those. If you decide to build a new system then consider the following: Size the case for future expansion. Typical full tower cases will fit about nine drives. You might need something fancier if you want to go bigger. Make sure the mobo has at least two M2 slots. Those work well for cache drives.

I was reading some more about this and realized that the components in the drive can seize up over time. Then I remembered that I have some old CD-ROM drives where the tray got stuck because the drive belt wore out. But even in those cases you can replace the broken component. Same with hard drives, especially considering that recovery companies do just that, replace worn out or broken components so that you can get your data. Anyway, I guess my point is that the story doesn't quite add up. Even a hard drive sitting in storage could probably be recovered, especially if the contents are valuable.