zeus83

Any errors in the system log?

Hi, check this thread for starters: If display shows up an unraid logo during vm boot then the gpu is passing through. However if it shows nothing on windows boot it might be a driver issue. When I had 6800 xt I had much more troubles with it than with nvidia cards.

Hi, first add this to your vm configuration: <timer name='tsc' present='yes' mode='native'/> On my gaming VM important sections look like this. Don't ask me why it is so. I've spent years of making a perfect gaming VM 🙂 <features> <acpi/> <apic/> <hyperv mode='custom'> <relaxed state='on'/> <vapic state='on'/> <spinlocks state='on' retries='8191'/> <vpindex state='on'/> <synic state='on'/> <stimer state='on'/> <reset state='on'/> <vendor_id state='on' value='1234567890ab'/> <frequencies state='on'/> </hyperv> <kvm> <hidden state='on'/> </kvm> <vmport state='off'/> <ioapic driver='kvm'/> </features> <cpu mode='host-passthrough' check='none' migratable='off'> <topology sockets='1' dies='1' cores='8' threads='2'/> <cache mode='passthrough'/> </cpu> <clock offset='localtime'> <timer name='hypervclock' present='yes'/> <timer name='hpet' present='no'/> <timer name='tsc' present='yes' mode='native'/> </clock> Also check tsc time is working correctly on host: Check your system log and if you see something like: x86/split lock detection: #AC: CPU 0/KVM/17187 took a split_lock trap at address: 0xfffff80556c4f75d Then try solution from here:

Hi, you might want to try this: sysctl -w kernel.split_lock_mitigate=0 or add the boot parameter: split_lock_detect=off You can read about the issue here: https://pve.proxmox.com/wiki/Split_lock_detection I've recently faced similar issue when running Division 2.

Yeah sorry, I've just recently upgraded my second unraid server to 6.12 and before I had a pool with device by-id names. It seemed at first that unraid kept the original naming, but looks this was not true.

Hi all, I've rebooted my server and zfs no longer imports pool referencing devices by-id for unknown reason: May 23 22:22:52 chia-unraid emhttpd: mounting /mnt/zarray May 23 22:22:52 chia-unraid emhttpd: shcmd (37): mkdir -p /mnt/zarray May 23 22:22:52 chia-unraid emhttpd: shcmd (38): /usr/sbin/zpool import -f -N -o autoexpand=on -d /dev/sdc1 -d /dev/sdd1 -d /dev/sde1 -d /dev/sdg1 -d /dev/sdf1 -d /dev/sdh1 346450085342583949 zarray May 23 22:22:56 chia-unraid emhttpd: shcmd (39): /usr/sbin/zfs set mountpoint=/mnt/zarray zarray May 23 22:23:01 chia-unraid SysDrivers: SysDrivers Build Complete May 23 22:23:03 chia-unraid emhttpd: shcmd (40): /usr/sbin/zfs set atime=off zarray May 23 22:23:03 chia-unraid emhttpd: shcmd (41): /usr/sbin/zfs mount zarray May 23 22:23:04 chia-unraid emhttpd: shcmd (42): /usr/sbin/zpool set autotrim=off zarray May 23 22:23:04 chia-unraid emhttpd: shcmd (43): /usr/sbin/zfs set compression=off zarray May 23 22:23:04 chia-unraid emhttpd: /mnt/zarray root profile: raidz1 May 23 22:23:04 chia-unraid emhttpd: /mnt/zarray root groups: 2 May 23 22:23:04 chia-unraid emhttpd: /mnt/zarray root width: 3 May 23 22:23:04 chia-unraid emhttpd: /usr/sbin/zpool status -PL zarray 2>&1 May 23 22:23:04 chia-unraid emhttpd: pool: zarray May 23 22:23:04 chia-unraid emhttpd: state: ONLINE May 23 22:23:04 chia-unraid emhttpd: scan: scrub repaired 0B in 05:43:28 with 0 errors on Sun May 19 10:13:33 2024 May 23 22:23:04 chia-unraid emhttpd: config: May 23 22:23:04 chia-unraid emhttpd: #011NAME STATE READ WRITE CKSUM May 23 22:23:04 chia-unraid emhttpd: #011zarray ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 raidz1-0 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 /dev/sde1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 /dev/sdg1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 /dev/sdf1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 raidz1-1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 /dev/sdc1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 /dev/sdd1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: #011 /dev/sdh1 ONLINE 0 0 0 May 23 22:23:04 chia-unraid emhttpd: errors: No known data errors Previously devices in the pool were referenced by it's id like ata-WDC_WD40EFZX-68AWUN0_WD-WX62D329UVT0 and etc. Any idea how to fix this? I so used to see my pool status with normal device names... chia-unraid-diagnostics-20240523-1939.zip

Hi, I recommend you first get yourself familiar with this article: https://unraid.net/blog/zfs-guide Basically it all depends on your needs and use cases. raidz zfs is a good choice when you favour more disk space (that's what typically you need from NAS I assume) instead of IOPs. One advise regarding enabling zfs auto trim. You must be aware that auto trim doesn't come for free and may impact your pool latency and therefore performance. How bad the impact will be depends on the SSD controller and it's firmware. Some controllers drop performance significantly when doing trim and the latency will be increase in any case. Some others are doing well. So you need to test you SSD first to check your disk controller can handle this fine. Use command zpool iostat -vly 3 1000 <your pool name> To monitor your pool stats. You'll see the TRIM timings from there. So if there timings are bad or latency is critical for you then you should consider using a scheduled TRIM instead of auto trim. How often you need to TRIM your drives depends on your write workload. So it may be once a week or daily.

Hi, I owned lots of GTX 1080 and none of them was incompatible. I'd say with GTX 1080 / GTX 1080 Ti I had zero issues on passthrough. Did you follow the instructions on GPU passthrough precisely ? Do you have these lines in your VM configuration ? <features> ... <hyperv mode='custom'> ... <vendor_id state='on' value='1234567890ab'/> ... </hyperv> <kvm> <hidden state='on'/> </kvm> ... </features> Also if this the only GPU in your system try putting this onto boot line: video=efifb:off

Hi, 1) Yes, one of the options you to consider is to pass through SSD/NVME drive directly to the VM and use it for games storage. It is a nice options you'll get close to bare metal performance of the given drive. I've been using such option for years with my Samsung NVME drive with zero issues. The drawback is that this drive is not protected and in case of drive failure you'll loose your data. Which I think in case of games is not critical. Anyway you get what you get on the bare metal too. 2) Another option is to use BTRFS/ZFS pool and create a raw image drive for keeping your games. I would actually split your VM img and your game drive img for easier management. With such option you'll get your disk protection (in case of mirror or raidz setup), you 'll get instant snapshots , compression, l2arc caching (at least with ZFS options) and many nice to have 'enterprise' features. But you'll pay for it with your drive capacity (50% in case of mirror setup, ~33% in case of raidz of 3 disks). Performance wise the things will get complicated and you might expect slightly degraded performance, the same performance or even better performance. This all depends on your setup and tuning... And how much hours you want to spend on this. 3) Using an array for gaming vm is also viable option, because the array gives you a lot of cheap hdd capacity. However regarding the performance I would expect it will be the worst of all options. All options will work out of the box for you. But since you're new to this, I'd recommed the first one. It is a no brainer.

I wish I saw this topic earlier but anyway I will put this here, may be it'll help someone. Whenever you see your gpu under utilized the issue might be is the timer. Your VM 'clock' section must look like this: <clock offset='localtime'> <timer name='hypervclock' present='yes'/> <timer name='hpet' present='no'/> <timer name='tsc' present='yes' mode='native'/> </clock> Also read this [SOLVED] GEARS OF WAR 4/5 WINDOWS VM TERRIBLE PERFORMANCE In a nutshell you must ensure your host 'TSC' timer works correctly.

+1 Played Fortnite in VM for years with zero issues until recent update. Upset... Seems there is not workaround yet.

Great news on unraid support for ZFS pools! I've just recently migrated all my servers to ZFS with the help of community plugins. Therefore I think system level support for ZFS is a good starter and a right direction in further unraid development.

Has anyone had a success in running RTX 4090 on Windows 11 VM ? Whenever I run Windows 11 VM for some period the screen goes black eventually and never returns. The os seems to be running fine meanwhile, but no picture at all. Running the same setup in Windows 10 VM works just perfectly fine. And I tried RTX 4090 on my bare metal Windows 11 installation and there were troubles with running games, they simply crash soon after start. But screen didn't turn black.

Well this explains why autostart works fine on my registered unraid and doesn't on trial one. Thanks!

I have the same issue to me it seems. My unraid on 6.11.3 doesn't autostart array even I set so.

Yes, I only have 6600 xt in my disposal right now. Works pretty well in VM. However I set up everything you did, but once rbar enabled in bios it's black screen when I start my VM. No any errors in the logs. But I've noticed also that VM manager stops after I shutdown the VM. May be there is some critical issue with that.

How did you manage to do this ? Whenever I enable resizable bar support in bios my VM starts with black screen, no video output at all on my 6600 XT.

Resizable bar is not virtualized so far. It's disabled in QEMU: https://github.com/qemu/qemu/commit/3412d8ec9810b819f8b79e8e0c6b87217c876e32

The overclocking won't work in this container. Since the only way to overclock nvidia gpu in linux is using nvidia-settings which requires X server running. Which is not the case in this docker. And it's not trivial thing to setup (at least for me :D) You may try following this guide:

Thanks a lot for this.

Hi, I've prepared the instruction in a way that you don't have to see this QXL paravirtual graphic card. If you still see it in the VM then you do something wrong. If you share your VM xml I can be more specific.

Hi, this thing helped me. Ensure you have 8444 port forwarded and then change network type in chia docker template from Host (as recommended) to Bridge. I have zero sync issues after I did this.

I've just built a local image in 5 minutes , no time to chill 🙂

The docker builds chia-blockchain inside, and it takes the latest by default if I got it correctly. But for some reason they 're not preparing new docker images. Therefore I think the only way is to build docker image manually.

They've only added tzdata package, nothing important. And anyway it won't build like that way they did it. I've tried to build docker image by myself from 1.1.6 tag specifically but the sync still doesn't work in linux docker version. On windows GUI it works perfectly. *UPDATE* I've changed template network type from Host to Bridge and it has started to sync immediately. I was able to sync 3k of blocks in less than 5 minutes.