JonathanM

That statement on its own would be enough for me to return the drive, considering the forces needed to make that dent INSIDE a shipping container... however... obviously changes things. If you could return it and get perfect drive in return for nothing more than hassle on your part, personally I might return it. But given that the drive seems to be performing, maybe it's not worth the aggravation. How long after accepting and using the drive would you be able to return it without loss to you? I'd definitely be keeping an eye on it.

You could set up a temporary trial install of Unraid to boot the Ubuntu machine and see if it mounts.

Sure, just set a new config to remove all the disks you want to erase from the array, and run the preclear script on them. A normal preclear is very effective at erasing disks, it would take a lot of money and time to get any data off a properly precleared disk.

Reformatting is super quick, and totally ineffective at really wiping data. All it does is write a new table of contents and show all addresses as available for writing new files, nothing is done with existing data beyond removing the easy to access table of contents. The closest thing to a DOD wipe would be preclear with random data before it zeroes it. It is extremely time consuming, but any method that repeatedly writes values to every single address on the disk is going to take a very long time, no way to shortcut a thorough wipe.

I understand that, but if it mounts ok in Unraid using Unassigned devices, you could look in the syslog and use the mount options that worked there.

googling custom-cont-init.d seems to point to a ls.io container, which neither krusader or handbrake are. How long have those extra shares been around?

Depends on your perspective. If you only use usenet, or only seed torrents for bare minimum, then there is nothing wrong with setting it up the way he does. It's only been in the past few years that linking files in user shares has worked properly, so it wasn't even an option to optimize with links. People set up their systems years ago with what was best practice at the time. Setting up a new system from scratch allows many more options nowadays, it's a little harsh to call old accepted practices completely wrong when they still work, just not as optimized as using the newly available features. Maybe you need to set up guides to help people migrate to the newer setup without losing data?

I wouldn't call it better, but it works for me and has for many many years, and inertia typically wins. apcupsd is NOT affiliated with the APC company, so YMMV. Since you are familiar with and use NUT, you are probably better off sticking with it. Open source support of proprietary hardware is always a little hit or miss.

Have you had issues using apcupsd in client mode? I've been running with my Unraid server as the apcupsd master and all my other physical machines and VM's using apcupsd slaved to the Unraid server. It's worked great for me, everything starts an orderly shutdown staggered between 1 to 5 minutes after the power is out, and everything has been properly put to bed 15 minutes into the outage.

Cool, I just keyed on the 25 minute runtime being good, which if you shut everything down promptly is fine, but if you are expecting your IT closet to stay functional for 25 minutes that would be an issue. Many folks assume a battery backup is meant to continue running as long as possible during an outage, which only applies to multi kilobuck setups that typically have their own rack, and usually have a diesel generator that steps in after a few minutes. Consumer type battery backups are meant to get things shut down safely if the power is out more than a minute or two.

Do you have ports forwarded on your router or passed through on your endpoint firewall? If so, which ports are you pointing them to on your Unraid box, and which containers or apps are responding? Are those containers or apps up to date and currently widely used across the internet? Are they configured according to their app specific best practices? AFAIK, network security isn't a black and white thing, if you open services to the outside world, you have to stay current and monitor those services to be sure they are as secure as possible. BTW, this is a general practices type of question, not really Unraid specific.

Try 6.11rc3 and see if it alters this behaviour.

I'd call that barely adequate. Applying several assumptions and rules of thumb, first being that you don't want to drain a typical UPS below 50% to prolong eventual battery replacement, so that means less than 15 minutes of usable runtime, and depending on how long it takes your typically running VM's to properly shutdown when asked, you could be running on the ragged edge of getting everything shut down properly even if you call for shutdown at power out +5 minutes. I'm not saying you necessarily need more capacity, just that you need to manage what you have, and make sure your VM's get the message to start shutdown pretty much immediately on power out, via a properly configured apcupsd network client in the VM. If using windows VM's, make sure they are set to hibernate on power loss, as shutdown may trigger a lengthy queued windows update.

Does it mount ok in Unassigned Devices in Unraid? Something is strange with your setup, Unraid doesn't use mdadm at all, and to my knowledge a valid XFS drive in Unraid would never be set up as anything other than a single normal partition across the whole drive. A valid drive in Unraid should mount with no issues in any linux that supports modern XFS.

Only unclean shutdowns trigger checks, so avoid unintentional shutdowns by using a properly configured UPS and verifying that the server can successfully shutdown unattended. If you can't get that done, then be sure to set the array NOT to auto start, so you can better control the bootup sequence after an event. When you shut down normally, stop the array before you power down, that way if something is hanging up you can investigate the cause and get the array properly stopped so you won't get an unclean shutdown.

No firsthand experience, just a couple google hits. https://xfs.org/index.php/XFS_FAQ#Q:_Does_the_filesystem_have_an_undelete_capability.3F https://github.com/ianka/xfs_undelete A quote from that page... Doesn't sound very hopeful.

@SpencerJ?

Linux is case sensitive, ARTIST and Artist are two different items, and can coexist without a problem. SMB as configured in Unraid to avoid other issues is case insensitive, those two items are seen as identical, so only the first occurrence is shown, any duplicates are hidden.

2 data disks, 2 parity, Maybe if the parity is still valid from before the incident there may be a chance?

Yeah, restore from backup is the only real option. XFS recovery is bad enough, add encryption to the mix and I'm not sure I would even know a good path forward. Encryption complicates data recovery immensely, pretty much on purpose. If you use encryption a comprehensive backup strategy is essential. WAY different. XFS recovery isn't exactly easy and the tools aren't free AFAIK. It's assumed if you are advanced enough to use XFS with encryption, you know enough to keep good backups.

The built in memtest is very old, it's the latest version licensed for distribution from a foreign source (foreign meaning not directly downloaded by the end user from memtest86.com) It's incompatible with UEFI and many newer board features. You will either need to temporarily disable UEFI boot in your CMOS (not ideal) or download the latest free version from the memtest site and create a boot USB on another stick.

Not supported Parity array and SSDs are a poor fit, no trim, and other issues Not sure what you mean here, pools can be configured with BTRFS RAID levels, but it's not recommended to mix HDD and SSD in the same pool. Maybe possible, but still not optimal. Normally HDD are used in the parity array, with the largest set as the parity volume, and others as data volumes. SSD's are used in pools, either single per pool or multiple in BTRFS RAID sets, or some combination of pools that works for you. Given your stated goals, the closest suggestion I can give is set the 12TB as disk1 in the parity array, with no parity disk, and the SSD's in a BTRFS RAID pool. I'm not sure of the stability of BTRFS RAID5, I would be more comfortable recommending RAID10. You could set up backup software in the VM clients and point them to a share on disk1 of the array.

How, exactly, did you accomplish this? Depending on how you set things up, it's possible that you are using the VPN directly instead of going through the privoxy proxy server.

Since you mention swapping PSU's, I'm dropping a reminder to never use modular cables from a different PSU unless you use a meter to verify correct connections. Incompatible cables that appear to connect the same can permanently fry expensive parts.

Click on the container on the dashboard, select console cd /data dir touch test.txt dir rm test.txt dir The first dir should show the contents of /mnt/user/download/incomplete The second dir command should show the added 0 byte file test.txt, the last dir show just the original folder content. You should also be able to see the test.txt appear in the share if you refresh the content of the folder in any share view.