testdasi

Is there any harm to use these high values for relatively smaller arrays (e.g. let's say 5-disk 30TB)?

Great vid! I was super doped realising that my first slot is connected to node 2. That means I still can leave unRAID to do unRAID things in node 0 like it prefers and have my workstation VM isolated on node 2. Now if only I can force it to only use the RAM connected to node 2. You can always flip the cards and pass them through via vbios method. For gaming VM, it's better to lock everything to the same node (although you are running in UMA not NUMA so not sure how much benefit it's gonna get). How do you that please? It's always a "hope and pray" situation with me to get RAM assigned correctly. Now that I isolated the entire node 2 for my VM, it would allocate 99% of my RAM to node 2, which is good (enough) - but ONLY after a clean reboot. The longer I wait after the reboot, the more it allocates to node 0 (sometimes it would go nut and allocate 100% to node 0, even though there's nothing else running). Don't know how to do it without rebooting.

Apparently there's already commands to tell which core is on which die. @bastl @Jcloud Perhaps you guys can try to see what shows up? ~# numactl --hardware available: 4 nodes (0-3) node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 node 0 size: 48208 MB node 0 free: 350 MB node 1 cpus: 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 node 1 size: 0 MB node 1 free: 0 MB node 2 cpus: 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 node 2 size: 48354 MB node 2 free: 4680 MB node 3 cpus: 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 node 3 size: 0 MB node 3 free: 0 MB node distances: node 0 1 2 3 0: 10 16 16 16 1: 16 10 16 16 2: 16 16 10 16 3: 16 16 16 10 Apparently can even check which VM is using how much RAM connecting to which node: ~# numastat qemu Per-node process memory usage (in MBs) PID Node 0 Node 1 Node 2 ----------------------- --------------- --------------- --------------- 33117 (qemu-system-x86) 1751.71 0.00 2442.32 33297 (qemu-system-x86) 2840.03 0.00 1326.58 82938 (qemu-system-x86) 28445.78 0.00 20757.30 91591 (qemu-system-x86) 182.21 0.00 8052.15 ----------------------- --------------- --------------- --------------- Total 33219.73 0.00 32578.35 PID Node 3 Total ----------------------- --------------- --------------- 33117 (qemu-system-x86) 0.00 4194.02 33297 (qemu-system-x86) 0.00 4166.61 82938 (qemu-system-x86) 0.00 49203.09 91591 (qemu-system-x86) 0.00 8234.37 ----------------------- --------------- --------------- Total 0.00 65798.09

Problems: The case has 8 expansion slots, your 4 GPU will occupy all 8. You won't have space for the USB card. You either need a new case or some creative modding. The 2nd GPU will cover the middle PCIe slot so you will need creative use of PCIe extender(s) to make it work. The Taichi X399 middle slot is PCIe x1 (albeit with open end) so I'm not sure your 4-controller PCIe x4 USB card is going to work in that slot. Theoretically it will just be slower. Theoretically, you need a case with at least 10 expansion slots: GPU1 (x2) - extender in - USB - GPU2 (x2) - GPU3 (x2) - GPU4 via exender out (x2). It's still not going to be easy to (a) get an extender to stretch over 5+ slots over 2 big GPUs and (b) your GPU3 + 4 will completely cover all the ports at the bottom of the board so access is going to be a massive pain. Also perhaps consider a Gigabyte board since it has full-length middle slot. I would also recommend opting for compact GPUs but it looks like you guys are reusing your existing stuff so that's not an option then. You might want to think simpler. There's no need for a separate USB controller if hot-plugging isn't a requirement. You can pass through individual USB devices to unRAID. If you all use exactly the same model of peripherals, it's going to be a massive pain to identify things and edit xml but should still work. The motherboard has 2 separate USB 3.0 controllers that can be pass-through to VM (in addition to a shared USB 3.1 controller that can be used for individual USB devices pass-through). So if you can live with just having 2VMs with dedicated controllers and 2 VMs with no hot-plug (preferably the 2 with distinctly different peripherals) then that simplifies things. In short, take the USB controller out of the question and the build might just work.

Do you know which AGESA was the proper fix in? That probably helps the TR peeps to know for sure the min BIOS to use.

Memory interleaving may be the difference because it relates to the Threadripper design. A Threadripper CPU is essentially equivalent to a dual-CPU / quad-CPU in the server world, which leads to the UMA / NUMA distinction. When the CPU is in UMA mode, memory is interleaved and exposed to both dies with priority for throughput. When in NUMA mode, there's no interleaving and each die access its own memory bus first and then the other die i.e. priority for better latency. In other words, UMA treats the CPU as one unit and NUMA treats each die as its own CPU. For the 1950X, UMA / NUMA can be selected. For the 2990WX, for the same reasons that you mentioned, only NUMA mode is available. So when it comes to pairing logical cores to physical cores, it might be done incorrectly in UMA if the numbering is based on NUMA. It also makes sense why the 2990WX has a different numbering scheme since NUMA is the only option. Of course, that's just my hypothesis since I can't turn on interleaving on my 2990WX to test.

So I updated to the latest template. Is there anyway to pass additional parameters to openvpn? My idea is to use remote-random to allow the docker to pick a random server at every restart. The section of code below is deleted every time the docker start and replace with the server in the template. Only the remote-random line remains so I'm guessing something was set up to remove any lines starting with "remote " remote-random remote de-berlin.privateinternetaccess.com 1197 remote de-frankfurt.privateinternetaccess.com 1197 remote czech.privateinternetaccess.com 1197 remote france.privateinternetaccess.com 1197 remote ro.privateinternetaccess.com 1197 remote spain.privateinternetaccess.com 1197 remote swiss.privateinternetaccess.com 1197 remote sweden.privateinternetaccess.com 1197

@bastl: One thing I can think of - did you have memory interleaving on or off? Unfortunately I can only wish that I have a 1950X laying around. I aint Linus.

So it seems on the 2990WX, the numbering already follows what @thenonsense said (e.g. 0 + 1 are a pair - see attachment). Wonder if it's a BIOS thing since only latest BIOS can run Threadripper 2.

Thanks a lot guys. With 2990WX it's gonna be even MORE complicated between the "fast" vs "slow" cores.

Try one of these: Download latest driver of your network card, uninstall current driver, restart, reinstall driver. Use regedit to delete all entries of any memorized network (wired and wireless), restart. Reinstall Windows from scratch The fact that your other VM / machine can access unRAID fine suggests your unRAID config is good. So all we can do is to try to force Windows to "forget" what it had that didn't work. All have worked for me at one point or another. Why (3)? I have had a weird corrupted windows update (the major type - one that creates a windows.old folder) that suddenly caused things to inexplicably fail which was only fixed by reinstalling and reupdating from scratch.

I think there's a bug. The VPN parameters are not passed to the docker. It ends up using the opvn file in the appdata folder. So if you want to change the server to anything other than the default nl PIA server, just edit the opvn file.

Are you planning to update to BIOS 3.20? Note in case you missed it: Asrock said must update to 2.30 first before going to 3.20.

Do you have the 2GPU the same way you have in the pic? i.e. GTX 1060 in first (primary) slot and HD5450 in 4th (secondary) slot? I have never had 2 different GPU in the same machine so have never managed to figure it out. Does unRAID only use the GPU in the primary slot? If the primary GPU is passed to a VM, does unRAID automatically switch to using the secondary GPU? Otherwise then basically the secondary GPU only purpose is to extract the rom of the primary one. I'm kinda reluctant to waste the primary PCIe slot on a cheapo GPU just for unRAID - especially when there are PCIe x1 GPU out there. Would be nice to be able to mount a 1070 on a true x16 slot + a cheapo GPU on the x1 slot and leaving both x8 slots + 1 true x16 slot for other purposes.

Cheers mate. Looks like I should go Tai Chi direction with my build too. For your 2 GPUs, did you have 1 for unRAID and the 1060 for VM? or both passed to VM? The Samsung B-die RAM thingie, that only relates to running overclocked right? I plan to run my RAM at stock but just want to make sure. Did you add multifunction option to your syslinux acs override?

@binhex: I think the VPN has to support port-forwarding in order for the web ui to work. Tested with several PIA servers and noticed this. Wonder if it's worth putting that on the first page in case people need to know. This is the official list of PIA servers supporting port-forwarding: CA Toronto CA Montreal Netherlands Switzerland Sweden France Germany Romania Israel

Yes, that would be awesome!!!

Squid: please reinstate Sort By Date, I beg thee. That's the only way I know to check out the latest new apps without the old ones clogging up the view. E.g. if I last checked on 20/06 and found "duplicati" as the newest one, then if I check today 28/06 then I automatically know only to check out apps appearing above "duplicati". Otherwise without the Sort By Date, the "New / Updated" view has 45 apps, for which perhaps just 2-3 are really new since my last check. It is very hard to find 2-3 needles in the haystack. And like bags with the missus, there are apps I don't even know I need until I see them - IF I manage to see them.

Does this need to be done every time unRAID is restarted? And why 6.3.0+ only? Was it a new undocumented feature or something?

Would this work? Install teamviewer on a VM. Use TeamViewer from a remote computer to control said VM. Run browser remotely through TeamViewer and access the GUI.

Well, connecting from outside can be done through OpenVPN and buying a product without proper research is a waste of money...

Emaling feature doesn't seem to work.

Do it through the console (e.g. launch Midnight Commander aka "mc") or Dockers (e.g. Krusaders / Dolphins). You MUST remember to only move files from DISK to DISK. Moving files between share and disk is an easy way to lose it. If you are not sure of what you are doing then I recommend to just leave it.