Pinch

Members
  • Posts

    84
  • Joined

  • Last visited

Converted

  • Gender
    Undisclosed

Pinch's Achievements

Rookie

Rookie (2/14)

0

Reputation

  1. I have been struggling to get gpu passthrough working for some time now, but it just won't work on Unraid. When ESXi is installed on the system, the gf 1070 will pass through fine to a Windows guest, but the same on Unraid will result in code 43 -- the install is a virtual machine. Hyper-V has been set to 'no', but despite of this Windows still clearly states the system is a virtual machine in the task manager*. Nothing's hidden from the guest it seems. Edit: Yeah, turns out Hyper-V should be set to 'yes'. Still didn't fix the problem, but finally got an error in the log. The hardware: Gigabyte Aorus Pro X570 + Asus ROG Strix X570-E Gaming (tried both; same problem) Ryzen X3950 EVGA 1070 I have tried all the xml additions googling help with code 43 will give to no avail. Since the OS is saying it's a virtual machine, I'm not suprised by the Nvidia drivers being aware too... -- this must be the issue, right? Booting without a VNC gpu will give a black screen. ROM has been dumped and set up. Does anyone have any idea why Windows won't get fooled? I am fairly desperate at this point. *On the 'Performance' tab. If the system's not virtual, it will state whether it is virtualization-capable or not.
  2. Has anyone successfully built the custom drivers? I tried twice in two different environments, but got the same script error in both cases. I don't currently have any of the test environments running, so I am unable to provide the exact error -- for now, I'm just curious if anyone's actually managed to build the drivers using the provided scripts?
  3. Sorry, of course: /usr/local/emhttp/plugins/dynamix.docker.manager/scripts/docker run -d --name='openvpn-as' --net='bridge' -e TZ="Europe/Paris" -e HOST_OS="unRAID" -e 'PGID'='100' -e 'PUID'='99' -p '943:943/tcp' -p '9443:9443/tcp' -p '1194:1194/udp' -v '/mnt/user/appdata/openvpn-as':'/config':'rw' --cap-add=NET_ADMIN 'linuxserver/openvpn-as' Edit: It's fixed, though now idea why. I reinstalled openvpn-as and it threw the "ERROR: Cannot ioctl TUNSETIFF as0t0: Device or resource busy (errno=16)'" in the webui when trying to start the server. Changing nothing, but perhaps trying to start the server a few more times, it was suddenly running and working. It's working, but I don't think the problem is solved.
  4. I am completely stumped by this: My OpenVPN-AS has been running perfectly and then out of nowhere stopped working. When it stopped working I was still able to connect on the TCP port, but after restarting the container a few times, that stopped working too. The unraid server has been restarted to no avail. The logs showed errors pointing to the interface or ports being already in use, which made no sense to me... After having tinkered with some settings (after it stopped working -- setting the container to privileged mode and only listen to UDP on port 1194 in the web ui), I am not seeing any errors in the log, but the server still doesn't listen on the UDP port. Can anyone explain to me what this means: root@Tower:~# netstat -an | grep 943 tcp6 0 0 :::943 :::* LISTEN root@Tower:~# netstat -an | grep 1194 udp6 0 0 :::1194 :::* The web ui (943) is working fine, telnet'ing to 1194 on localhost results in connection refused. UNIX is not my strong suit; what am I to gather from the fact that it doesn't appear to 'LISTEN' on port 1194? If I switch off privileged mode, the errors are back: 2019-08-07T09:22:20+0200 [stdout#info] OVPN 0 ERR: 'Wed Aug 7 09:22:20 2019 ERROR: Cannot ioctl TUNSETIFF as0t0: Device or resource busy (errno=16)' 2019-08-07T09:22:20+0200 [stdout#info] Server Agent initialization status: 2019-08-07T09:22:20+0200 [stdout#info] { 2019-08-07T09:22:20+0200 [stdout#info] "errors": { 2019-08-07T09:22:20+0200 [stdout#info] "iptables_web": [ 2019-08-07T09:22:20+0200 [stdout#info] [ 2019-08-07T09:22:20+0200 [stdout#info] "error", 2019-08-07T09:22:20+0200 [stdout#info] "service failed to start due to unresolved dependencies: set(['web'])" 2019-08-07T09:22:20+0200 [stdout#info] ] 2019-08-07T09:22:20+0200 [stdout#info] ], 2019-08-07T09:22:20+0200 [stdout#info] "openvpn_0": [ 2019-08-07T09:22:20+0200 [stdout#info] [ 2019-08-07T09:22:20+0200 [stdout#info] "error", 2019-08-07T09:22:20+0200 [stdout#info] "process started and then immediately exited: ['Wed Aug 7 09:22:20 2019 ERROR: Cannot ioctl TUNSETIFF as0t0: Device or resource busy (errno=16)']" 2019-08-07T09:22:20+0200 [stdout#info] ], 2019-08-07T09:22:20+0200 [stdout#info] [ 2019-08-07T09:22:20+0200 [stdout#info] "error", 2019-08-07T09:22:20+0200 [stdout#info] "service failed to start or returned error status" 2019-08-07T09:22:20+0200 [stdout#info] ] 2019-08-07T09:22:20+0200 [stdout#info] ], 2019-08-07T09:22:20+0200 [stdout#info] "web": [ 2019-08-07T09:22:20+0200 [stdout#info] [ 2019-08-07T09:22:20+0200 [stdout#info] "error", 2019-08-07T09:22:20+0200 [stdout#info] "process started and then immediately exited: 2019-08-07T09:22:18+0200 [stderr#error] twisted.internet.error.CannotListenError: Couldn't listen on 0.0.0.0:943: [Errno 98] Address already in use." 2019-08-07T09:22:20+0200 [stdout#info] ], 2019-08-07T09:22:20+0200 [stdout#info] [ 2019-08-07T09:22:20+0200 [stdout#info] "error", 2019-08-07T09:22:20+0200 [stdout#info] "service failed to start or returned error status" 2019-08-07T09:22:20+0200 [stdout#info] ] 2019-08-07T09:22:20+0200 [stdout#info] ] 2019-08-07T09:22:20+0200 [stdout#info] }, Netstat is showing the same thing as before. The web ui is loading, but I am unable to sign in using the correct credentials. What happened? Why did this stop working all of the sudden?
  5. I am trying to do the same thing: An Unraid 24/7 file server running a workstation VM with passthrough GPU that is only powered on when needed, but so far the GPU is not powering down when the VM is in any way. Is the answer to this question so obvious it doesn't even merit a response?
  6. Success! Flashing the M1015 on the ASRock resulted in the PAL error, but my trusted old unraid tower with an Asus mb did it without issues. So the M1015 is flashed as an LSI 9211-IT with the latest P17 fw and pass-through is working! I chose to flash the BIOS too, although I have no intention of connecting a boot disk to the controller - is longer boot time really the only drawback of flashing the BIOS? If yes, that's not a problem. An odd thing is that when connecting an old Toshiba 160GB 2.5'' to the M1015 it won't recognize it. It works fine when connecting it to the onboard ASM1062. I have no intention of actually using this disk, but am still wondering why it won't work... I couldn't find a definitive answer for this one: will drive spin-down work? ATM I have a spinner and an SSD connected to the M1015 being passed through; both drives are apparently spun down, which doesn't make sense - my old tower would never show an SSD as spun down, so I don't know if I can trust unraid on this one... EDIT: Silly me, the SSD in the old tower is the cache disk with a lot of plugins running, so even if it was a spinner it would never be spun down - is it normal for unraid to show an SSD as spun down when it's idle?
  7. According to jaddel's link I think I got it wrong: LSI9240 apparently IS the M1015, which can be flashed to an LSI9211. Thanks for the link, jaddel
  8. Hmm, seems like I'm either really unlucky that the ASRock card won't pass-through or there's something wrong with my setup. It doesn't make sense 5.0 doesn't recognize any of the controllers... I took the chance and ordered an M1015... As far as cross flashing goes, I get a lot of hits where people's been flashing it as a LSI-9240-8i, but you mentioned LSI-9211-8i. Any specific reason you're using that controller? Do you happen to have a link of some sort regarding the cross flashing?
  9. I meant the setup utility wouldn't even recognize any controller, so there was no disk available to install to (had one disk hooked up to each controller). 5.5 supports only the Lynx Point controller (don't know about the Marvell controller) 5.1u1 supports both the Lynx Point, ASM1062 and the 88SE9120, but can't pass-through the Marvell 5.0 apparently supports none
  10. Thought I'd give 5.0 a quick try, just to see if it would pass-through the Marvell controller; the initial setup wouldn't recognize any of the three connected controllers, so I think I'm gonna stick with 5.1u1
  11. Think I'll just order the IBM M1015 then... Just to be clear: this card will pass-through (at least using ESXi 5.0), right?
  12. Thanks Keeping this thread on subject, I just moved an ASRock SATA3 (Marvell 88SE9128 / 88SE9120) from the old tower to check if ESXi would recognize it. It does and disks are read fine. But when passing it through to either unRAID or a Win7 VM, neither can see the controller - Win7 doesn't even show anything unrecognized in the system management. I am planning on replacing this controller either way (it only has two ports), but is it odd that ESXi can use it, but apparently not pass it through? Will a different - confirmed working - controller most likely still work?
  13. I have no need for the added features in 5.5, so as long as I'm not the only one having issues, I'll have faith it'll be fixed at some point... If only I knew why plugins aren't being automatically installed, I would be ready to migrate to the new server. It's not really an issue since they can be installed manually, I just hate not knowing why it's not working!
  14. While on the subject of ESXi and unRAID, what's your experience with performance? I am currently not able get more than 50-60MB/s on parity checks - doesn't matter if I only have virtual disks mapped or if only disks on the pass-throughed controller are mapped. Another weird thing is that plugins (.plg) located in /boot/config/plugins won't install automatically. If I installplg them manually they work, also after a reboot... I'm close to being ready to move my current array from my old server, but am still a bit hesitant because of the performance issue and the plugins not installing (it's such a basic feature, makes me think I've set up something incorrect). I'm using the pre-built VMDK from http://lime-technology.com/forum/index.php?topic=26639.0.
  15. So I tried both resetting the BIOS and at the same time installed ESXi 5.1u1 instead of 5.5 - much to my own surprise pass-through worked! Tried reinstalling 5.5, not changing any BIOS setting, and it was broken again, so it appears 5.5 is the issue... I'd like to know why 5.5 fails, but at the moment 5.1u1 works, so I'm happy