May 6, 20179 yr Hi, hope someone can help. On Windows 10 VM creation, I see the following error: internal error: process exited while connecting to monitor: 2017-05-06T18:47:37.955194Z qemu-system-x86_64: -device vfio-pci,host=0b:00.0,id=hostdev0,bus=pci.0,addr=0x6: vfio: failed to set iommu for container: Operation not permitted 2017-05-06T18:47:37.955252Z qemu-system-x86_64: -device vfio-pci,host=0b:00.0,id=hostdev0,bus=pci.0,addr=0x6: vfio: failed to setup container for group 16 2017-05-06T18:47:37.955262Z qemu-system-x86_64: -device vfio-pci,host=0b:00.0,id=hostdev0,bus=pci.0,addr=0x6: vfio: failed to get group 16 2017-05-06T18:47:37.955289Z qemu-system-x86_64: -device vfio-pci,host=0b:00.0,id=hostdev0,bus=pci.0,addr=0x6: Device initialization failed The following is set: BIOS: OVFMHyper-V: No (as it's Nvidia, though I've not yet seen why it's recommended to be off) The groups for the two cards are:IOMMU group 16 [10de:128b] 0b:00.0 VGA compatible controller: NVIDIA Corporation GK208 [GeForce GT 710B] (rev a1) [10de:0e0f] 0b:00.1 Audio device: NVIDIA Corporation GK208 HDMI/DP Audio Controller (rev a1) IOMMU group 19 [10de:128b] 05:00.0 VGA compatible controller: NVIDIA Corporation GK208 [GeForce GT 710B] (rev a1) [10de:0e0f] 05:00.1 Audio device: NVIDIA Corporation GK208 HDMI/DP Audio Controller (rev a1) Selecting both the video and audio for VM creation on either the card in group 16 or 19 generates the above error. There is nothing else in these groups, and the cards are not being used by another other VM or Docker. Each card is intended to be used in separate VM's. I've tried the Enable PCIe ACS Override, which didn't make any difference to the error, as I expected since the groups seem fine. I've also tried the following in syslinux.cfg: label unRAID OS menu default kernel /bzimage append vfio_iommu_type1.allow_unsafe_interrupts=1 initrd=/bzroot The cards are both NVIDIA GT 710, one Asus and one MSI, which I'm sure that I read someplace that 700 series upwards should be fine with UNRAID. Does anyone else have any ideas?
May 16, 20179 yr Author Hi, _jonte: I've looking pass through the cards, so from what I understand SeaBIOS is out of the question. 1812: I did a full reinstall and removed one of the NVIDIA cards, just in case that there was some clash. Same issue though, so no luck on that front. Please see the attached diagnostics, hopefully, you'll spot something that I've missed. tower-diagnostics-20170514-1123.zip
May 16, 20179 yr The cards are fine, i have several 710's. you need to put this back in your syslinux.cfg, as it does not show up in your system log: append vfio_iommu_type1.allow_unsafe_interrupts=1 initrd=/bzroot then reboot, and try again. You have a proliant, this should resolve the issue. If you get another error, post it and diagnostics again. Don't reboot and post "fresh" diagnostics as it is missing information that may be helpful.
May 16, 20179 yr Author Thanks for the quick response. I tried your suggestion previously (see original post), it's missing from the last diagnostics since it was a re-install. I've re-applied the configuration and am still getting the same error, but for group 19 as you'd expect as that's where the 710 is. Still no luck though. Please see the attached new diagnostics. Thanks. grid-diagnostics-20170516-0647.zip
May 16, 20179 yr 2 hours ago, dorian said: Thanks for the quick response. I tried your suggestion previously (see original post), it's missing from the last diagnostics since it was a re-install. I've re-applied the configuration and am still getting the same error, but for group 19 as you'd expect as that's where the 710 is. Still no luck though. Please see the attached new diagnostics. Thanks. grid-diagnostics-20170516-0647.zip Quick question while I look through the logs: did you specify onboard video as primary on your server bios? doesn't matter, read next post. Edited May 16, 20179 yr by 1812
May 16, 20179 yr This is your problem: May 16 06:46:30 Tower kernel: vfio-pci 0000:05:00.1: Device is ineligible for IOMMU domain attach due to platform RMRR requirement. Contact your platform vendor. This is the reason/workaround: You may be able to blacklist the card and get it to use audio and video, but I haven't fooled with that yet. Edited May 16, 20179 yr by 1812
May 16, 20179 yr Author I've never had to set a video card for the primary BIOS, so haven't looked as explicitly stating that it be used. At this moment there are two video outputs: 1) Onboard ATI VGA, which is connected to a screen to I can see the lower login, and the server posting. 2) NVIDIA 710, which is HDMI. Given that fact that the VGA is outputting the post and tower login, I'd assumed that the onboard ATI would not be an issue.
May 16, 20179 yr Author Thanks...... I'll play around with that tonight and see what can be done. It just seems that UNRAID and HP servers don't like to play nice together. Guessing it'll be Dell next time.
May 16, 20179 yr 2 minutes ago, dorian said: Thanks...... I'll play around with that tonight and see what can be done. It just seems that UNRAID and HP servers don't like to play nice together. Guessing it'll be Dell next time. There are certainly quirks to using software that wasn't intended to be used on enterprise servers actually on enterprise servers.
Archived
This topic is now archived and is closed to further replies.