SelfSD

It's past the 60% mark now and I'll post the diagnostics after everything is working so you can get the whole story @ChatNoir. Don't worry, I won't restart the server between changes. @JorgeB Maybe that is worth a footnote in the post since raid1cX is now an option in Unraid?

I gave it a try. The pool is up with 1 disk less and fully accessible but it did not start a balance or device delete. It's just repeating "Warning - Cache pool BTRFS missing device(s)". The log says the following: Aug 19 07:35:13 UNAS kernel: BTRFS warning (device dm-5): devid 3 uuid xxxx is missing. Aug 19 07:35:14 UNAS emhttpd: shcmd (851638): /sbin/btrfs device delete missing /mnt/pool_name & So Unraid attempted to remove the missing device, but it was not able to. I got no errors besides the device missing. I'll make a bug report about it once I have all of this down. I attempted to do a manual device delete and it gave me the following error: "error removing device 'missing': unable to go below three devices on raid1c3". I'm now converting it back to raid1 and will report back once it's done.

I'm about to remove a drive from a 3 disk raid1c3 pool to move it to the array, but the pool is encrypted. While reading the FAQ on how to remove a cache pool disk, this part got me confused: Does this only count for non-redundant pools or are all encrypted pools SOL when it comes to removing drives? Unraid 6.9.2

You can safely stop it and read the data that has been copied so far. If you set the target to /dev/sdX, it's as simple as mounting it like normal. You may need to reboot or reconnect the drive for UD to see the new partition table. You may also need to run a checkdisk on it from a Windows PC if it won't mount properly. Just note that if you did not specify a mapfile, you can't resume the copy process if you wish to attempt to copy the bad sectors later. It will have to start over from the beginning.

According to the Network UPS Tools documentation some APC models, including the one you have purchased are rebranded Microsol Solis UPSes so it does not work with the APC driver and from what you are writing, it does not appear to have usbhid support either. You should be able to use the NUT plugin with the "solis" driver. Just remember to set "Start APC UPS daemon" to "No" or else NUT might not start properly. Just note that they are saying the following: Sources: https://networkupstools.org/docs/man/apcsmart.html https://networkupstools.org/docs/man/solis.html

I never solved it. Turned everything I could off, back on, back off and so on. No change. With 6.9 since Beta 25 it became less frequent like you are seeing but still happened. Now I'm on a new build so I can't comment further on the issue unfortunately.

6.9.0 stable makes it possible to write to the array at gigabit speed but reading from the array is still just as slow. Does anyone else use AllwaySync, GoodSync or Free File Sync between a Windows PC and their Unraid server?

Hi guys! I'm currently on 6.9.0-RC2 and I've been trying out a couple of different backup tools in hopes of getting faster backup speeds, but all of them (AllwaySync, GoodSync, Free File Sync) were limited to about 30 MB/s copying to and from the array. Now the funny part is that I can run 3 backup sets in parallel to utilize my network fully, but that doesn't help much for single files that are rather large. Copying files to and from my array using Windows explorer does so at full gigabit speeds. Win10 Pro 20H2. What could cause this? No single core on my computer or the server are maxed out during transfers. I have done some reading before posting this about SMB tuning and other performance tweaks. I have only enabled Direct IO in Global Share Settings which did not improve the transfer speeds in any of those programs. Enabling SMB Multi Channel did not help either. If anyone has any ideas I'd love to hear them!

No problem! I have been running 6.9.0 beta 30 for just over a week now without any problems. I had some issues on 6.8.3 and 6.9.0 beta 29 which seems to have all been solved in beta 30 as it has been rock solid with an uptime of almost 9 days now, which is a new record for my old AMD system in a long while now lol.

That's exactly your problem problem. Driver version 440.59 is not supported after PMS 1.20.2. You need at least driver version 450.51 on linux. If you don't want to update to a beta release (thanks @MowMdown 😄 Didn't notice that he had a non-beta prebuilt!) head on over to the link below and get the "Unraid Custom nVidia builtin v6.8.3" version and copy the 8 files from the ZIP onto your flash drive. Don't forget to make a flash backup before doing so! You might have a new GPU UUID after upgrading but you can still get that from the LSIO Nvidia plugin page, or grab the Unraid Kernel Helper and it will show there too.

So I have absolutely no idea where to start reporting this problem or what to give out as I can't see anything in the Plex debug logs, docker logs or Unraid logs. I thought it might be a disk I/O issue but it's not. Maybe someone else can test this and confirm if it's a isolated issue for me or something that's more widespread. When running the LSIO Plex container on a separate VLAN, it struggles to send out the next buffer segment to any devices on the LAN, causing the video to stop and wait until it catches up. Sometimes it manages to send out the next segment before the playback catches up but it does not It appears to work fine outside of my own network. VLANs are controlled by a Unifi Dream Machine Pro, running through a MikroTik S+RJ10 with the VLAN passed through and 10Gbit on all ports to a Mellanox ConnectX-3 on my server. Unraid 6.9.0 beta 30

Heads up to anyone running 6.8.3. Plex Media Server was just updated to 1.20.2 on the public side, so your GPU accelerated hardware encoding will stop working if you update. @linuxserver.io's beta 29 and @ich777's custom beta 30 build has a newer driver that is compatible with this latest release of Plex Media Server.

I just got another reply from another user who said the following: If this is true everyone who uses the Nvidia Unraid build is SOL once 1.20.2 of Plex drops for the public. ☹️ Edit: Just confirmed this on Nvidia's websites: https://developer.nvidia.com/nvidia-video-codec-sdk/download

I posted in the Plex forums and have gotten some replies so far. In 1.20.2 the required API version was updated. Public release is 1.20.1 at the moment and currently works. I just got a reply from someone who ran driver version 440.100 and had the same issue running Ubuntu Server on bare metal. For them upgrading to 450.66 fixed the issue. Should I post in the Unraid Nvidia topic too in case it requires a driver update to work? It would really suck if NVENC hardware encoding completely stops working when 1.20.2 is released to the public. ☹️

@saarg Ah. I downgraded to the latest public build and HW transcoding is working again. I'll forward the issue to Plex. @mgutt If only I could do it that easily on Unraid 😭

Hardware encoding seems to be broken in one of the newest container image updates. The Plex server believes the Nvidia driver is out of date and will not do hardware encoding. Sep 28, 2020 06:59:19.216 [0x152ceebf5700] ERROR - [Transcode] [FFMPEG] - The minimum required Nvidia driver for nvenc is 418.30 or newer Sep 28, 2020 06:59:19.217 [0x152cec7e3700] ERROR - [Transcode] [FFMPEG] - Driver does not support the required nvenc API version. Required: 10.0 Found: 9.1 I'm on Nvidia edition 6.8.3 with the 440.59 driver.

Just upgraded from 6.6.7 and everything looks good on first boot!

The quickest way would be to only transfer the world over to the existing server. If you don't have a network share to the appdata folder, set it up and give yourself read and write access. First stop the server from the web interface if it's running. Go to 'appdata > minecraft > servers > NameOfServer > world' and delete the files and folders in there. Then open your existing server world folder and copy (Do not move the files! Always keep a copy in case something goes wrong) the files and folders into the new location. Now you can start the server and log in to it and your kids old world should be there.

That sounds more right. It's been a while since I had a MineOS docker installed.

"minecraft" all lowercase if I remember correctly.