TimTheSettler

Can you elaborate on the following? How many HDDs did you have in the array before? What size of drives were they? Where are they now? How many HDDs do you have as your backup drives? What size of drives are these? How many SSDs did you buy? What size of drives are these? How many drives can you physically fit into the case? How many SATA ports do you have (how many drives can you connect to the system)? How much data are the hard drives holding right now (in total - across the board)?

Wow, it's been ages since I heard of a daughter card. Had one in my old 386. I'm always worried about switching from one system to another without some kind of backup in place. What I mean is that I would do one of the three following things with #1 being preferred and then #2 and finally #3 as a last resort. Buy some new hard drives and build a new system. Copy the data from one to the other so that you have two identical systems (from a data perspective). Use syncthing to keep both systems in sync. Or use Duplicati or Vorta to use one system as a backup for the other. Buy a couple new hard drives and build a new system. Copy the data from the old system to the new one. When a hard drive becomes empty in the old system (because the data was copied to the new system) then add it to the new system so that you have more space. Gradually all the data will "move" from the old to the new but everything will be protected by the parity drives. Unplug the unRAID USB from the old system along with the hard drives and plug those things into the new system. Away you go. Your new system will start up with the same settings (and hard drive mapping) as the old system.

One thing that I came across as a difference was the LAN chipset. I bought a motherboard with Intel I219V and I originally installed TrueNas. I then read that it's better to have the I219LM for something like TrueNAS because the OS was server-grade. In fact the spec specifically says, "Server operating system support" among a few other features. But, I have no idea what any of that really means. In the end my onboard I219V seems to be doing just fine with unRAID. https://www.intel.com/content/www/us/en/content-details/333229/intel-ethernet-connection-i219-product-brief.html?wapkw=I219LM vs I219V Maybe someone here can shed some light on the difference is between a desktop LAN chipset and a server LAN chipset.

I've moved drives around and I'm pretty sure that it uses the model + serial number of the drive to identify it. You should be able to take apart the computer and then re-assemble it and, assuming you have the original configuration, it should be able to just pick up where you left off. If you don't have the original configuration then, as Decto states, you'll need to identify the drives back into their proper place.

I use Vorta. I also use the Wireguard VPN in unRAID to connect the computers over the internet.

I wanted to write up an article on "Backup Strategy". As I state below, when you talk about "backups" you should really be talking about a backup strategy and "copying files" or "using parity" is only a part of that strategy. This is a work in progress and I welcome suggestions, critiques, and comments. Definition A person or thing that can be called on if necessary; a reserve. Computing: the procedure for making extra copies of data in case the original is lost or damaged. Why is data backed up? A backup is created to protect the original by duplicating it so a backup strategy should involve protection so as to avoid the need of creating, maintaining, and having a backup in the first place. How can files be lost or damaged? They can be lost by being stolen digitally (hacked or ransomware) or physically (computer or hard drive burgled). They can also be lost due to negligence (accidental deletion) or on purpose (virus or ransomware). They can be damaged by time (bit rot), deficient hardware (drive failure), or by physical means (fire or other “act of God” event). Backup Strategy When people talk about “backup” they mean a backup strategy. Copying your files to another location is the most simple form of backup but it’s really a part of an overall strategy. There are two main parts of a backup strategy and these are further split up into other areas. Protection Encryption Access Quality Hardware Location (Offsite/Lockdown) Redundancy RAID/Mirror/Parity Copies/Snapshots/Archive Prevention and Recovery As you develop your strategy it evolves into a more effective method of prevention and more consideration is put into the ease of recovery.

Is that good or bad...?

I'm not sure if this will help or not but this is what I've done. Let's say that we have two servers and on each server there's a share/folder called "documents". Use syncthing to synchronize the files in the "documents" folder. Any changes on one server would be synchronized to the other. Then, using Duplicacy, set up an archive somewhere on one of the servers and tell Duplicacy to back up the "documents" folder to that archive. I use Vorta/Borg and it runs each day taking a snapshot each day. If there are no changes then the archive remains the same size. If I move some files around in the folder that is being archived then the archive stays the same size because the file contents haven't changed but the archive records the change of location. The daily snapshot gives me a history of file movements if I need to track that. If a file was deleted or changed unexpectedly then I can recover it at any point. This setup solves the following problems or provides these features: Off-site backup. If any server goes down you can recover from the other server. Multiple synchronized file servers. Backup archive: Incremental recovery. File change history. Efficiency from deduplication.

Awesome video. Thanks man!

The cache drive article is good but I would say the following: The cache drive serves two main purposes. It provides storage of things that need good performance like docker images. As the article above states, having 50GB to 200GB is good for this need. It provides a buffer for transferring files. Transfer across network Since the mover runs every hour it's good to have 450GB for this (1000Mbps * 60 seconds * 60 minutes / 8 = 450GB/hr). Internal file transfer is another story but how often will that happen? So 500GB is good but barely. 1TB is much nicer and 2TB is great. The article concurs. How many cache drives are needed? I would say two because the cache is basically mirrored. So two is better than one. Do you need three or four? Not really but you could use the extra drives to add to the first cache (and its mirror) or you can add a second cache (one cache for dockers and such and the other cache for file transfers).

Single is better than multi? I thought the idea was to have dedicated rails for dedicated power?

It's cool that the Z790 actually support 8 SATA ports (https://www.intel.ca/content/www/ca/en/products/sku/229721/intel-z790-chipset/specifications.html). The motherboard uses a Realtek LAN which many will tell you to stay away from. So far I've been lucky (?) since it's worked for me on my one server but I might be one of the lucky few. Generally it won't work because the Linux driver has problems. I once installed OPNsense on a mini-PC using Realtek and it kept crashing. The motherboard has 4 M2 slots which is also pretty cool. I see that you plan to get 4 cards for those. Consider bumping up the size since 1TB and 2TB are good prices (bang for your buck). Some people really like ASRock. Give them a try. CPU: Are you trying for efficiency or for capability. Make sure it can do virtualization and has transcoding capabilities. Read about that in other places here on this forum. I would argue that it's the hardware features that you should be really interested in. You have an odd set of hard drives. I'm assuming that you have these lying around somewhere already otherwise I would just buy some 18TB drives (best bang for your buck). Even if you get three of them you're farther ahead.

I think he's right about the advantages: Parity implementation allows for any size of disk to be added to the array. Can be installed on almost any hardware allowing any type of system to become a server. Not limited to custom hardware enclosures/solutions. Intuitive web-based interface. Impressive support of and choice of docker containers and VMs. Not so sure about all his negatives: Performance due to way parity is implemented. Maybe in the grand scheme of things but if it's mainly a server that you access through your network to store files or watch movies then performance is great. Cost. This one is odd because he doesn't like the license cost (I tend to agree) but earlier he said that it was "cost effective" when you considered the fact that you're not tied to particular hardware. Lack of deduplication or compression. This depends on the file system you use. If you use ZFS then you can do both although deduplication is not worth it. If you use btrfs then you can do compression. Lack of enterprise support like LDAP. Lack of hardware support.

Why not keep the old NAS and use the new one as the main unit. The old NAS becomes your backup. Use Vorta or Duplicati or whatever backup software you want to back up the data to the other NAS. Obviously you'll have limited space so you just need to decide what gets backed up. At some point you might want to just convert that whole thing into another unRAID machine.

It might seem disheartening when a feature is added to a product and only a few people use it at first but I think it's one of those "look-ahead" things that gives me confidence in unRAID. If I never see improvements, be they features or bug fixes, then I get worried about the future of the product. Keep up the good work and please keep adding new things here and there.

I don't know about you guys but all questions like this one, where you ask is this better than that, seem to be answered as a comparison and then "depends on your needs". Has anyone ever got, "Oh yeah, you need to use this one for sure otherwise you're an idiot for choosing that other thing."

Ouch. ChatGPT misspelled set up. It bothers me that setup is used as a verb when it's a noun. XFS is easier to manage and setup set up than ZFS, but lacks some of the advanced data protection features of ZFS. And while we're on the subject. This is wrong too. It should be "Shut down the system".

Same with me. Never had power problems but I've had SATA cables that "break" when re-seating the connectors. I can only guess that the SATA cable was crappy and moving it around broke a wire in the cable. Problem: Hard drive not recognized but is getting power. Solution: New SATA cable. This happened recently to me after a power failure. The server was on a UPS. There are shutdown log entries showing that the server was shutting down when the UPS hit the low-level warning. Maybe it ran out of power just before the final shutoff. Not sure since I wasn't there. Problem: Server wouldn't start due to weird error. Solution: "Repair" flash drive in Windows and then restore from the latest backup. Server came back up without any problems. I have a similar problem. For me right now it seems to be a tray/motor problem. I just use a pin to force the tray open. Once it's open then the tray will open and close like normal for a bunch of times and then, out of the blue, the tray will get stuck again. I've come across a number of fan-related problems. Usually they are because the fan is old or gummed-up and accumulated dust. An old server I had would make some god-awful noise when it started up but if you smack the fan housing then the fan re-seats itself and never makes noise again unless it's shut off and turns back on. Noisy fans due to age. They accumulate dust. If you don't clean them regularly then the dust adheres to the fan over time and makes the fan spin in erratic ways. New fan worked for about a month then failed. It caused the server to turn off. I had it plugged into a fan connector that told the motherboard to turn off the computer if the fan RPM speed dropped to a low level. This was designed to avoid an overheating CPU. It's a cool feature but not when you don't know about it. Took me hours to figure out that the failed fan was the cause of the server shutdown.

Here's a historical benchmark for you. You'll notice that I replaced the two 18TB drives with two 14TB drives (hence the reason for the parity sync of the two 14TB drives on Sat and Sun in the middle). Note that one of the parity checks took four days. If the server is in use during a parity check then the check will take a lot longer. These are all Exos drives.

Add the following to your SMB settings. https://forums.unraid.net/topic/84988-smb-settings-to-force-nobodyusers/

That's how I understand it. The daemon is restarted so that a new log can be started.

When drives appear, disappear, and then reappear it could mean the drive is dying but could also mean that the cable is broken or not connected properly. I've had this happen to me in some older systems I've had. It could be that I played with the cable a few too many times and I ruined the wires or it was a crappy cable to begin with. Once I replaced the cable the drive came back and ran beautifully for a few more years. I've also had cases where the drive appears to be getting errors and then I move it around to another drive bay or put it into another computer and then everything is ok. The errors (usually bad sectors) are re-evaluated and are not bad sectors after all. There seems to be no rhyme or reason sometimes. Thermal shock and excessive heat are what usually kills components. My oldest working hard drive (shown below) has been running continuously for over 12 years.

@BobNik @adminmat @ConnerVT @nraygun Hi guys. I was curious about this myself and here are some things that I found. As you can guess the apcupsd daemon controls the UPS communication. The number in brackets is the process ID of that daemon process. It seems that, on a weekly basis, the old process [524] is destroyed (exited) and a new process [13221] is created. Obviously your process numbers would be different from mine. This weekly restart process is controlled by "logrotate". You can find logrotate.conf in the "/etc" folder. Basically the logrotate config file is run and then all scripts within the logrotate.d folder are run. In the /etc/logrotate.d folder you'll see apcupsd which contains the following commands. The rc.d is a folder that contains "run commands" for daemons. More info here --> https://linux.die.net/man/8/logrotate and https://forums.unraid.net/topic/119894-resolved-problems-with-apcupsd-starting-at-regular-times-when-disabled/ The UPS Self Test is a test that the UPS performs on a regular basis and the apcupsd daemon simply posts the results to the log. It's interesting my settings show SELFTEST = NO and yet my UPS still does a regular Self Test. Maybe just a bug with apcupsd. See this for more info about the Self Test --> https://www.apc.com/ca/en/faqs/FA405317/ Note below in the screenshots that everything in my settings jives with the logs especially the power failure at 6:50:34 lasting for 2 seconds.

"Belongs to them" has a few meanings. Any data that's "out there" is "out there" which means that it's sitting outside of your network so Mr. Grey is partially correct simply because someone else has it. What they do with it is another thing. If Google or some other site says that it's your data then that's all well and good but they still have your data and you need to trust that their network doesn't get hacked and then your data is no longer yours alone. This is why I like the idea of using a cloud service but having the data encrypted which is what apandey is doing. Even if the backup site is hacked your encrypted data has no meaning to the hackers.

I wouldn't bother if you're just a typical home user which many people here are. However, there are some nice benefits that @itimpi has mentioned. Bit rot detection is nice but a rare occurrence and hard drives are designed to catch this so is it worth it just for that? Maybe. ZFS also has snapshot but would you use that? No. The one thing that I like is that ZFS supports inline compression such that when you save a file it's compressed and then saved to the disk in the compressed state. This can save some space for you. However, you need to keep in mind that reading and writing files requires the extra step of compressing and decompressing it. Will unRAID support that feature? I'm not sure. Another cool thing is deduplication but this requires massive amounts of memory and some planning of file block sizes since the file block determines how much memory is used to keep track of all the blocks. There are also ways to improve performance like ARC/L2ARC and SLOG/ZIL. Are you confused yet? Join the club. It's not an easy file system to just plug in and use. So when you take all these "configurable" parts away you're left with "just another file system". So for now I don't see a benefit but as people use the file system and other smart people make it easier to use and/or set up then it will become "the next best thing". This is what I'm hoping for. I see the array or caches as being synonymous with a ZFS vdev. The drawback is that you need to think about your setup before you build your server whereas right now it's all pretty simple with unRAID which is what I love about unRAID. Forgive me but I think the following is a great explanation of ZFS. The link is to TrueNAS. Someone here needs to write up something similar but with reference to unRAID once ZFS officially shows up for unRAID. https://www.truenas.com/docs/references/zfsprimer/