bastl

from the BOINC wiki: https://boinc.berkeley.edu/wiki/GPU_computing Maybe this project isn't supporting GPU computations. Drivers are up-to-date for me and I have my VM with a 1080ti the BOINC client is detecting the GPU. Setting the client to use the GPU even when in use doesn't change anything. GPU load monitored with GPUZ doesn't ramp up. Not even a single spike.

Oh, I didn't noticed this thread yesterday. I helped a guy yesterday in another thread with the hint which version of q35 to use. I knew i had to change it to an older version in the past, but couldn't remember which was the max version that works. @toolmanz Maybe edit the first post with the solution, so people can find it easier.

@ogi Did you tried your GPU in another slot on the board? If not, try this. You might have to adjust the syslinux config. Some boards have issues passing through a GPU plugged into the first slot. Also adding a cheap GPU in the first slot and using a GPU in the second or third slot is an option. On my x399 board I have 5 pcie slots and having only 1 card in slot 1 won't work for my. By adding a second card I'am able to pass them through both, no matter which slot I use. Also keep in mind if you using a vbios in your xml for your card you use the right one for your card. You mostly find different revisions on TechPowerup. Maybe try a different vbios. If it's a Nvidia card you wanna passthrough you have to manual hex edit the vbios and remove some of the headers like SpacInvaderOne described in one of his videos.

You're hitting the max speeds of the SSDs. If you need more than 500MB/s you need NVME drives for example on both sides or storage that is able to handle more than that.

@Joeri1594 I tried to setup a Pfsense VM and got the same issue with one core stuck at 100% load. The only way i could setup Pfsense is to emulate a different CPU. Host Passthrough (default) of of the physical CPU won't work for me. Try to set the CPU Mode to "Emulated Qemu64" or manual edit the XML with the following. Don't forget to adjust the core count to match your settings. <cpu mode='custom' match='exact' check='full'> <model fallback='forbid'>Skylake-Client</model> <topology sockets='1' cores='2' threads='1'/> <feature policy='require' name='hypervisor'/> <feature policy='disable' name='pcid'/> <feature policy='disable' name='hle'/> <feature policy='disable' name='erms'/> <feature policy='disable' name='invpcid'/> <feature policy='disable' name='rtm'/> <feature policy='disable' name='mpx'/> <feature policy='disable' name='spec-ctrl'/> </cpu> Edit: Only tried with Q35 2.6. Newer versions won't work for me.

@Joeri1594 What I did with this tweak is trying to replicate the physical device which is a Quad port nic plugged into a single slot. Thats the reason why I adjusted the slots. This way the the VM should detect it as a single device in a single slot with different functions like the physical device is recognized. This shouldn't be a problem and as I said helped some users with other devices like GPUs.

This is a common error in case a device isn't reset from the VM. Users with AMD GPUs know that issue. Only a server restart helps in this case. This could also happen to none AMD devices if let's say a VM crashes or hangs and the passed through devices aren't reset propably or the device itself aren't able to reset. Maybe you can try to use the fix that helped some AMD users by putting all nics on the same bus with different functions. Adjust the xml with the following: <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x01' slot='0x00' function='0x0'/> </source> <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x01' slot='0x00' function='0x1'/> </source> <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x1'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x01' slot='0x00' function='0x2'/> </source> <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x2'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='yes'> <driver name='vfio'/> <source> <address domain='0x0000' bus='0x01' slot='0x00' function='0x3'/> </source> <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x3'/> </hostdev> I've put them all on the same bus (bus='0x01') in the "address type" line, same domain/slot and counted the function up starting from 0 (function='0x0'). Maybe this only works on q35 and not i440fx what you selected. Not sure. Try to setup a new VM with the Q35 template and the different addressing I've posted above. It's just an idea, maybe it helps.

@thamuz If you use a VNC GPU and a GPU passthrough at the same time, the virtual VNC GPU will always become primary and the passed through physical GPU will produce a code 43. For me it worked out best to first install Windows with VNC only, install the drivers from virtio iso, make sure network is working and finally setup RDP or other software like Anydesk or Teamviewer to remote manage the VM. Next step is to shutdown the VM, remove VNC and add the GPU and it's audio part. Before starting the VM you have to connect a monitor to the GPU. If not, the GPU won't initialize fully or you can end up in scenarios where it hangs during boot. If you see only one core with 100% load and it won't change after a couple minutes, reduze the core count of the VM to only 1 core and try again to boot the VM. Minimize the devices you wanna pass through to a minimum. Try it first with only the GPU, NO USB devices like keyboard or mouse, NO USB cards, NO physical network nics. Make sure the GPU is working and add the other devices later, one by one.

@SpaceInvaderOne First of all a big big thanks for this container and all your useful videos you made. Helped me a lot to get started with Unraid. Now to my main question. I today installed Macinabox for the first time and noticed something. There are 2 folders created in the appdata share. What's the reason for creating 2 shared folders, both empty? Is it supposed to be this way? I quickly pulled the High Sierra install files with your Docker and installed it on an UD drive. Install went fine so far. VM is up and running after adjusting all the paths in the xml.

Why is TRACKING enabled by default?

If you create a new VM on a AMD system with "host passthrough" selected, the cache mode will be automatically be added old xml: <cpu mode='host-passthrough' check='none'> <topology sockets='1' cores='4' threads='1'/> </cpu> new xml: <cpu mode='host-passthrough' check='none'> <topology sockets='1' cores='4' threads='1'/> <cache mode='passthrough'/> <feature policy='require' name='topoext'/> </cpu> For existing VMs you have to adjust it yourself.

@Squid After reading the inotify man page I doubled it to 1048576 (1G). Let's see if the warning is gone.

Thanks for the hint. It's currently set to the default of 524288. Can you explain what this setting is for and maybe why only disk 2 shows these warnings?

Update from 6.8.2 went fine so far. Unraid is back up runing with all VMs and Dockers. Small thing i noticed is by opening the "system logs" under tools the orange animated loading stripes showed up for split second and dissappeared. Not happened before and doesn't look like it has any impact. Maybe only a small glitch. Something that I already reported in the 6.8.2 release thread is some warning for my Intel nics during Unraid boot Mar 7 14:14:16 UNRAID kernel: igb 0000:04:00.0 eth1: mixed HW and IP checksum settings. Mar 7 14:14:16 UNRAID kernel: igb 0000:06:00.0 eth2: mixed HW and IP checksum settings. . . Mar 7 14:14:16 UNRAID kernel: igb 0000:04:00.0 eth0: mixed HW and IP checksum settings. Mar 7 14:14:16 UNRAID kernel: igb 0000:04:00.0 eth0: mixed HW and IP checksum settings. . Not sure how to interpret this warning. Looks like network has no issues so far. And another warning that still comes up since 6.8.2 is the following: Mar 7 14:27:13 UNRAID inotifywait[7814]: Failed to watch /mnt/disk2; upper limit on inotify watches reached! Mar 7 14:27:13 UNRAID inotifywait[7814]: Please increase the amount of inotify watches allowed per user via `/proc/sys/fs/inotify/max_user_watches'. Smart values for the disk are ok. If someone is able to help me to understand these warnings please feel free to explain it. Thanks! old post for 6.8.2 with logs:

Oops. I completly missed the wrong time. 2018-01-16 😂

That's the exact message I get accessing the page. Can you check the certificate for the page? Example for Firefox click the lock item and select connection details and "more information" at the bottom. You should see a new window pop up where you can select "show certificate". Check if the Amazon cerificate is valid for you. Are you able to access Amazon or "docker.io" in China?

@JTok Quick question, is it possible to create encrypted compressed backups?

Can you try to access "https://registry-1.docker.io/v2/" via a browser? You should see something like the following: or can you try to reach "docker.io"? If I check the certificate for the page it's registered by Amazon. Either one of them isn't reachable for you. Not exactly sure how SSR or your "virtual router" is setup, how it exactly works or where the limitations are, but I guess you know, that you propably not able to reach all global sites from China without some sort of VPN or Tor.

@Yang-LU I quickly tried to to pull the jellyfin/jellyfin docker from your first post and I had no issues installing it. Your Timezone is set to Asia/Shanghai so I guess your server is in China. The "certificate has expired or is not yet valid" might indicate you're not able to reach the source of the docker you try to pull. Not sure how many chinese users Unraid has and where the limitations are in China but for me it looks like you can't reach the site for the docker where it is hosted. from your docker logs: time="2018-01-16T14:04:20.388405528+08:00" level=error msg="Handler for POST /v1.40/images/create returned error: Get https://registry-1.docker.io/v2/: x509: certificate has expired or is not yet valid" time="2018-01-16T14:05:23.771327113+08:00" level=error msg="Handler for POST /v1.37/images/create returned error: Get https://registry-1.docker.io/v2/: net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)"

The Linuxserver NC Docker got an update sometime the last hours. I guess he isn't using your docker.

This is exactly what you see, when you start copying files to your server. The first couple gigs are cached with the RAM after that, the transfers are directly written to the array. Direct write to the array means, as soon as you write to a array disk, the same data is read from it at the same time to generate the parity data, what causes another write operation. Reading and writing on a single disk at the same time causes the speed of the drive dropping to half it's possible speeds. As mentioned before you have to calculate in some overhead on top. If you have a drive in this chain, whats slower than the others, it will reduce the speed of the other operations. Also keep in mind if all drives are connected to the same controller, there also could be a bottleneck. Writing and reading from a array disk + writing to 2 parity disks at the same time + maybe a VM/Docker doin some read/write operations on a disk connected to the same controller can slow things down. Best solution for 10G networking is to use a fast cache drive, that has enough free space to catch the transfered data. There is no other way to saturate your 10G speed.

@limetech Just stumbled across these info on Phoronix https://phoronix.com/scan.php?page=news_item&px=Intel-KVM-CVE-2020-2732 "Sent out as notice "FYI" were three patches for CVE-2020-2732. Those patches were already mailed in as part of KVM fixes targeting the current Linux 5.6 kernel cycle and quickly pulled in by Linus Torvalds. Linux 5.6 Git is now protected from CVE-2020-2732 and should be back-ported to stable kernels soon."

@torch2k I have a extra line for an alias in the xml. Not sure if you need it, but worth a try. <filesystem type='mount' accessmode='passthrough'> <source dir='/mnt/user/backups/'/> <target dir='backups'/> <alias name='fs0'/> <address type='pci' domain='0x0000' bus='0x06' slot='0x00' function='0x0'/> </filesystem>

I have the same board with a 1950x and have 2 GPUs installed. 1080ti in the frist slot, 1050ti in the 3rd slot. Both can be passed through to VMs. For the GPU in the first slot I need a vbios to pass it to a VM. You can find the BIOS on Techpowerup for your specific model or you dump it yourself directly from your card. You might have to modify the BIOS with an hex editor to remove some NVIDIA headers. Maybe the following helps with VBIOS: I never tried the "Unraid NVidia build", so can't really tell if there is something special you have to count in. This build is to have native GPU transcoding support for Dockers if I remember correctly. "Unbind" the GPU will break things if something else depends on it. If you wanna prevent Unraid from using a specific card, get the PCI IDS from the TOOLS >>> SYSTEM DEVICES Example: IOMMU group 49: [10de:1b06] 43:00.0 VGA compatible controller: NVIDIA Corporation GP102 [GeForce GTX 1080 Ti] (rev a1) IOMMU group 50: [10de:10ef] 43:00.1 Audio device: NVIDIA Corporation GP102 HDMI Audio Controller (rev a1) and put it in your syslinux config. Under MAIN click on flash and scroll down and you see the syslinux config. Add the IDs so it looks like the following and restart the server. kernel /bzimage append vfio-pci.ids=10de:1b06,10de:10ef isolcpus=8-15,24-31 pcie_acs_override=downstream,multifunction initrd=/bzroot Unraid won't initialize the device on next boot and you should be able to pass it through. The isolcpus and acs part you might not need. In my example the cores from 8-15 and their threads 24-31 (second die) are isolated so my main VM is the only one that has access to it. And the ACS patch I need to split up my IOMMU groups to get an specific USB controller separated in its own group for passthrough.

@Marshalleq Just an idea. Some routers have "nat reflection" dissabled if even configurable. At least Pfsense has it dissabled on default. What I mean by that? On default if you try to access a services availabe to the internet wie DNS from your local network Pfsense will block it. Maybe thats the reason why these containers aren't able to talk to each other via the domain name of the Nextcloud instance.