Thx And Bye

This is the default behavior for many routers (e.g. OpenWRT) if IPv6 is enabled and the router handles DNS forward. By default they then advertise their own local IPv4 and link-local/ULA IPv6 as DNS servers via DHCP. Usually to support DNS resolutions for your local devices. You wouldn't want to advertise a global DNS server via DHCP in this case.

Retested with 6.12.3 with a full zfs share and btrfs cache pool. Changed Status to Solved

Yes that is correct. I'll keep that pool btrfs then and will report back when this is fixed in a stable release.

I've completed the transfer to all zfs drives now and only have zfs drives in my array. Now the reported space in Windows makes absolutely no sense at all anymore. Installed capacity is 2x16TB and 2x4TB and Windows reports 104TB total space. (currently verifying parity after the removal of the two remaining xfs drives) boxofcare-diagnostics-20230708-1743.zip

Completely missed to create and attach the diagnostic. Sorry for that. boxofcare-diagnostics-20230703-1309.zip

Yes I have Nvidia Driver version 535.54.03 and the latest plexinc/pms-docker installed.

I have this issue on the stable release 6.12.0, I never had this issue prior with any 6.11.x release. Adding a valid nameserver to /etc/resolv.conf instantly fixes the issue. Diagnostics are attached. This issue should be at least be treated as Minor and not just as Annoyance imo. boxofcare-diagnostics-20230623-2255.zip

The output from docker ps -a.txtlooks rather normal to me. Here is the docker inspect 2eda680bcc88.txt but I can't see anything special about it either? Here is the log of the containers starting:container startup logs.txt

Yes I have CA Auto Update Applications installed. The error occurs on manual backups too though and the backup and auto update run at different times anyways. It's always the same container that has this problem (it starts just fine when doing so manually via the webUI). The container is part of a stack deployed via docker-compose (Compose Manager Plugin) and not touched by the auto update. I've also replaced my gzip with pigz for (much) faster compression, but it's likely unrelated as the creation and validation of the archives works just fine.

I'm always getting a failed backup but none of the archives failes to validate. One container failes to start because the image can't be deleted (?) though. But a container not starting correctly shouldn't fail the backup, right? backup.log

Would it be feasible to get a checkbox to run nvidia-persistenced after the driver install on system startup to keep the driver loaded? More information: https://docs.nvidia.com/deploy/driver-persistence/index.html

Yes, two of them have. plexinc/pms-docker has it build in and I've created my own that checks that my VPN container is properly connected.

For now I'm running a user script on a daily schedule as a workaround: #!/bin/bash /usr/local/bin/unraid-api/unraid-api restart

I'm on version 2021.10.12.1921 currently, so it should be the most recent one. But this issue is present at least for a few versions now.

Hey, so I've noticed on multiple occasions that the MyServers plugin will lose connection and then cause a high CPU usage. It also spawns a lot of "sh" subprocesses without ever successfully establishing a connection again. executing 'unraid-api restart' solves this for at least a day or two. No more spawned 'sh' childs anymore too. Should I just restart via a user script every 24 hours as a workaround or is this a more general issue that can be fixed in the plugin code? All else kept the same, after a restart of unraid-api my load average drops from around 1.60 1.75 1.89 to 0.38 0.76 1.36 after about 10 minutes or so. This is significant CPU load and thus power usage that is completely wasted. If there are any logs I can provide I'll happily do so. The system log doesn't show any anomalies tho.

No but you can support this plugin in its endeavor: [Plugin] Docker Compose Manager - Plugin Support - Unraid

I wish there would've been a section about ECC memory but overall, it's a really informative episode. Good job!

After removing the old files and creating a new export it shows "related export file is not present": But the files are present on the flash:

I too can't access my Nginx via IPv6 anymore after updating to UnRAID 6.9.2. Connecting via IPv4 still works fine. I had no problem with my config in 6.8.3, 6.9.0 and 6.9.1. Is there a way to fix it or is this something that will get fixed in a future UnRAID update? EDIT: I did some digging and came across this pull request that was merged in Docker 20.10.2: https://github.com/moby/libnetwork/pull/2604 As UnRAID 6.9.2 switched from an older version than 20.10.2 to 20.10.5 it also includes this change and as such IPv6 isn't forwarded by default anymore. There is a fix merged: https://github.com/moby/libnetwork/pull/2608 but it's not implemented in the 20.10.5 release. It's fixed with 20.10.6 tho, see release notes: https://docs.docker.com/engine/release-notes/

It works for me expect for one drive (UnRAID v6.9.2). I suspect it has to do with the identifier in the webUI having a dot that is substituted with a underscore in the smart-one.cfg. WebUI: Samsung_SSD_850_EVO_M.2_250GB_S33CNX0H506075N Conf : Samsung_SSD_850_EVO_M_2_250GB_S33CNX0H506075N It's the dot in "M.2" that seems to cause the problem as changing the underscore to a dot in the config makes the values appear in the WebUI too.

Nice tips, I just wish it would be easier to setup KeysFile authentication and disable password authentication for the SSH. Just placing your pupkey in the UI and setting a checkbox to disable password auth would be nice. I currently have it setup like ken-ji describes here. Then i edited PasswordAuthentication to "no". Also think about a secure by default approach with future updates. Why not force the user to set a secure password on first load? Why even make shares public by default? Why allow "guest" to access SMB shares by default? Why create a share for the flash in the first place? I get that some of those things make it more convenient, but imo convenience should not compromise security.

Restarting the api solved the issue for me and it connected properly. The "Support Thread" for the plugin links just to the general support forum, so I assumed that I had to post problems in here.

Hi, so I have a couble of issues with the new "My Servers" integration of unraid.net. My setup is: - Unraid 6.9.1 server dedicated hardware, default ports for WebUI changed to 8080 and 4433 (SSL), as 80 and 443 are used by a container. - pfSense 2.5.0 on dedicated hardware the DNS resolver has unraid.net excluded for the DNS rebind: server: private-domain: "unraid.net" The problems I had: The first error is when logging in, I get the error "Communication with [UnraidName] has timed out" Problems I solved: 1. In the menu it shows My Servers Error, guest doesn't have permission to access "servers". Why would I want guests to access my server? - Solved by entering "unraid-api restart" on the cli as per the solution from ljm42 2. When trying to provision a Unraid.net certificate I get the error that the router has DNS rebind enabled, unraid.net is explicitly allowed already and the url does resolve properly. - Solved by entering the xyz.unraid.net:8080 domain with port manually and trying to provision again The only other thing that is working is the flash-backup. Is this all due to using non-default ports? Or is there some other problem that I need to resolve? If anything more specific is required, I'll try to provide it.