darianf

hello i didn;t understand what is the problem you are encountering (or the "many problems"). i assume you cannot start the Vm properly, or you can start it only once, or it's unstable. few things to consider: 1) go over the pinned post with the CPU pinning. you will find do's and dont's. Try to use single CPU - cpu0 for that Windows VM, till installation completes. then once is fine, allocate more cpu's. sometimes you will require also to put it to cpu0, for updates of windows os to succeed (i know, these updates are very hard to control their occurence). 2) try to dump the rom for that GPU card. then add it's rom file in the xml. adding the rom solved an issue with my 550ti card, where the VM could only start once per unraid box restart. any consequent try would fail... see more here: http://lime-technology.com/forum/index.php?topic=52960.msg535568#msg535568 Go over also the OP from gridrunner to see the guide... good luck -d

i found something, will check when i'll reach home, to see what's inside my xml https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Virtualization_Administration_Guide/sect-Virtualization-Tips_and_tricks-Libvirt_Managed_Timers.html

ekim, same issue here, on my x99-m ws from asus i have 4 controlles: 2 controllers are internal, no physical usb sockets exists. 1 controller is having all the usb 2 and 3.0, including the front i/o usb, internal usb (a small socket on the mobo where you could plug a cable and take it out via a bracket at the back of the case) - in total i think there are about 10 USB's - there are 2 buses here, on same IOMMU which does not support reset. unRaid flash is plugged here. so not suitable for passthrough. 1 controller is 3.1 - i was able to passthrough this one to my Windows VM without issues (unfortunately it has only 2 plugs so i need an extra usb 3.0 hub) i was also surprised why so few controlles for so many ports! but apparently this is common for the motherboards. Only solution that i could find is to add USB pcie controller (e.g. PCIE x1) - so my next purchase will be one of these...

that's the same behaviour i got till i did the cpu pinning based on your guide, and i also did passthrough for the entire USB controller (onboard 3.1) to the Windows VM. After that it reduced a lot, and only sometimes i could notice there's some very mild mouse lag... but will try to reduce it to 5/10, and see if it makes it even better. thanks!

thanks dlandon. i was able to install the fix common problems plugin without other issues (probably after some unraid restart). I have 32 GB of RAM in my x99 work box. - 2 VM's with GPU passthrough, one for folding (3 GB RAM), another one for regular work (8GB RAM, maybe up to 12, if really required). - 3 VM's for development (4 GB RAM each) - ubuntu 16.04, no GPU passthrough - sometimes an ocasional test VM with 2-4 GB RAM, no intensive work... None are high I/O throughput to disk. This box is used for work and not for data storage (i have different box for this purpose - in my signature) What should be an optimal setting? should i reduce from default 10/20 to 5/10? or even less? thanks, -d

thanks squid, will try also this thoroughly. at certain point i think i changed also there, but not sure what if the results were relevant... but i just restarted the entire unraid box and now the time did not advanced in guest VM. now thinking if power off vs restart of unRaid makes a difference. will try more - this is a hard nut to crack especially with this inconsistency... I am afraid that this is a case of OS trying to modify the Bios clock as well, and different OS will act differently.

hi trurl, the same computer was used up till few days ago, with windows 10, directly from the ssd. no issues with the time / time zone. Not sure what is the risk of playing with the time / time zone - basically following the same steps i described in my previous post - but probably the issue can be replicated easily. btw, i just updated to 6.3.0, but same behavior was present in 6.2.4 as well.

hi squid I try to explain the steps i perform while setting up the time of unraid tower (settings>date and time). In red what's wrong, in green what is expect behavior. NTP: no time zone: utc+2 athens, bucharest, istanbul I set the time to the current time and save - immediately the time is updated to 1 hour less, by the unraid interface probably - some javascript maybe... Then i change the unraid time to account for this strange behavior, by adding 1 hour. Save => the time is again reduced by 1 hour, but it's accurate as per my current time. Then i enable NTP as Yes, time textbox gets grayed out and is accurate. Next step is to start the Windows 10 VM. In the guest, i have similar settings like: time zone - utc+2 Athens, Bucharest Set time automatically is ON Time is 1 hour extra. Then i perform these changes: 1) Set time automatically is set to OFF 2) change time to reflect current time 3) Set time automatically is set to ON Time is accurate. I restart the Windows VM - time remains accurate I restart the unraid box. unraid time remains accurate after restart. I start also the Windows VM. Guest time is extra 1 hour. I go in Windows Settings, Date & Time. I switch OFF, then ON the "Set time automatically" = > time is accurate for the session. But then, any restart of unraid will change the time in the VM by adding extra hour. next step i will do to understand more of this, I will leave uncorrected the guest time, but will perform several unraid box restarts, and see when it stops adding this extra hour nonsense...

hi guys, this issue drives me crazy... Somehow the unraid insists to reduce the time i set manually by 1 hour. I have to account for this nasty thing, by adding 1 hour extra. finally i manage to set it up like this: ntp - yes time zone - utc+2, for athens, bucharest, istanbul (btw, if i set it up 1-2 minutes off, after NTP enabled it does not correct this - i would expect that the time will be adjusted to the second, but it's not the case - is this some clue that NTP is not working?) I start the Windows 10 guest, guest time is ok. Restart the VM, guest time remains ok. I restart the tower entirely, unraid time remains ok, but Windows 10 guest VM advances by 1 hour. Will try to restart the tower several times, to see how much i can time travel!

Try to perform the intallation assigning only cpu 0. Version migh introduces features, fixes but alao bugs...

thanks dlandon, i already did some of that as part of the optimizations for the Windows VM to be more specific on the fine tuning, i changed only the networking part, but not the disk caching. I read the blog post explaining the various combinations for the disk cache, but i didn't want to adjust them yet, as i wasn't sure what values will give me an improvement. My usage of VM's is not extreme, so I left them as default, assuming it's good for the regular usage just a side node - "fix common plugin" did not work for me. installation hanged for 2-3 minutes, i stopped it, then consequent attempts to install resulted in a puzzle icon for the plugin and clicking on the icon lead to blank page; will try to find if it's a common problem with this plugin thanks!

hi I want to follow up on my previous posts and to share with the forum my experience while passing through the evga 550ti 1gb GPU. I'll explain some issues/symptoms and workarounds - hopefully this might help others to troubleshoot similar problems with old cards... 1) I failed to dump the rom - exact error while dumping: "cat: rom: Input/output error" This was regardless of 550ti being in use, not in use, VM never initialized, restarted, bind, unbind - and any possible combination of these - so i gave up trying 2) Passthrough while in second slot, without the rom in xml, works only once per unraid restart. After start/stop of vm with the 550ti passthrough, consequent attempts to start VM will keep failing - so there's a need to restart the unraid box 3) I managed to passthrough 550ti by using the relevant ROM from techpowerup, but it works only partially, because : - it never worked in first pcie slot - so i had to use second slot. When i tried to use it in first slot, the fan went 100% - abnormal for the card, as it never goes above 70-80% even on when 100% load - it does not display anything in TEXT mode - so some OS's will not install with this GPU passthrough - as a workaround, i'm installing the OS's with my other GPU (zotac 1060 mini 3g), configure video drivers, etc, then switch back to 550ti. Because of this issue, after starting the VM, there is nothing displayed on the screen till the login, where it switches to Graphic mode. - Windows 10 VM and Linux Mint 18.1 does not like this card being passthrough, so the trick above did not work at all. I tried with both ovmf and seabios, same result... i vaguely remember that also Elementary OS 0.4 didn't like this card, but i'm not 100% sure, maybe i need to dig more. Ubuntu Gnome 16.04, Kali 2016.2 and Fedora 25 work ok, either directly or by using this trick to swap the cards after drivers installed. gridrunner, if you will, please add in your OP a small link to this post, as a note to the 550ti rom - to save some frustration... Edit: 1) adding all info in one place: the GPU part number i have is 01G-P3-1556-PK and the exact relevant rom was found in techpowerup. I did not try other roms. 2) despite the small inconveniences, i did not encounter stability problems with this card passthrough via rom file. Once it's initialized in Graphic mode, all is ok!

hi lionceau, what was the solution? to undervolt? As much as possible, i would suggest to try dumping the rom yourself - this will ensure best results. I did so for my 1060 3gb zotac mini, and never encountered an issue. -d

how to solve basic issues like connectivity, in case the user cannot reach the server via the admin console/ssh... tutorial is here : in short, it's like this: 1) you need video cards for this procedure: GPU1 for unraid output, GPU2 - for which you want to dump the rom, in slot 2 (so not in slot 1, closest to the CPU) 2) you need to identify the device id for that GPU2 & execute the commands for rom dumping - you might need to initiate a VM first (start/stop), in order to 'instantiate' the rom - not sure if instantiate is the right term 3) save the rom in an unraid location, where you have acces 4) edit manually the xml of the vm (this needs to be repeated every time you modify the VM via gui), and add in certain location relevant for your GPU you want to passthrough, an xml element with the path to the rom dump file. for my 550ti, i substituted the steps 1 2 3 with downloading the rom from techpowerup, because 1 2 3 didn't work for me as expected, I was getting an i/o error. For newer cards, most probably the procedure will give the expected results, while the rom from techpowerup will not work (different sizes/content) -d

hi dlandon thank you for this extensive guide and all the explanations supporting it. i just tried it on a windows VM. I'm not yet sure if it's because of the pinning, placebo or the fact that i passthrough the entire usb controller hosting the unifying receiver - but it does "feels" snapier and without any lag... I wanted to know if there is a recommendation for optimizing also the Linux VM's (and not just Windows). I assume it's similar concept, but thinking if there's something that needs to be done differently from Windows... thanks -d

hello Joseph, good to know that changing slots does not pose threat to array/assignments some info on the speeds of pcie, per version and lane... http://www.tested.com/tech/457440-theoretical-vs-actual-bandwidth-pci-express-and-thunderbolt/ regarding this point, try to take rom dump from the GPU cards. see the video/written tutorials - gridrunner's video is very concise and explanatory if procedure does not work properly, try to take rom from techpowerup, based on exact model that you have - i did the same, since my 550ti could not dump the rom properly. This fixed the issue where GPU could only be initialized once, after restart of unraid. but still i have issues with text mode - certain OS's i have to install with newer card, then after drivers updated, switch to 550ti... ah - 8800gt, my first card - i believe it was running crysis with medium settings (800x600?) Regarding the speeds of transfer and your 3 options, it's not something i can answer unfortunately ... (so I encourage you to try)

hi saarg thanks for suggestion i saw also this kind of mod. it's doable, to cut the extra dvi connector (present in most 1060 6gb and above) it's even simpler for lower end cards (1060 3g and below), as they have dual slot bracket but without extra dvi - so only the metal bracket must be cut "in half" In addition, some cards are also coming as single bracket (no need to cut anything), but the shroud, fan/s and radiator's thickness make it double slot - i was thinking at modding a card like this as it would be simpler. but i am not sure if they are still not too thick, even after fan/shroud removal. no way to measure i see that there are rx460's single slot - while i would try amd 460 as it's enough for my needs, but my other card is nvidia. i swap the cards quite often in the VM's definitions and having same team (e.g. nvidia) makes it easy for the drivers, especially in linux.

hi guys This post is more like asking advice on a H/W mod - more about physical constraints rather than compatibility with unraid. But I though to ask here since probably many folks are facing this issue trying to cram many devices in their builds Can somebody recommend an Nvidia 10x GPU, that is dual fan, where the radiator is thin (thinnest will be "suitablest" ). I intend to remove the plastic cover and fan 1 (the one closest to the rear I/O) with the hope to fit a short length device in next PCIE slot (e.g. a usb controller). Any ideas how to find such GPU card? -d PS: you cannot find these days single slot GPU's, and i feel the companies are missing a huge opportunity especially with the unraid comunity 1050, 1050ti are powerfull, cheap and modern cards that run cool enough to require small cooling solution - so why not?

initially i faced strange issue while setting the time in the tower. NTP as not getting correct time. when setting it manually, the time was update to a value +1 from what i was configuring. Had to do a trick to update manually to an incorrect hour, to account the for unraid's stubbornness to add the extra hour. Once the time was manually ok and time zone ok, i switched to automatically/ntp since then the time was ok in unraid. but now i see in windows vm, that the time advances every time i restart (could not yet identify if VM or unraid box restart) anything you can suggest please?

i would suggest to run headless (without video output for the tower unraid) as temporary solution till somebody can suggest for this particular issue... it means that you will pass-through even the slot 1 gpu, by adding it's rom in the vm xml gridrunner has a nice video tutorial on this process (how to obtain the rom and how to configure) -d

for the sake of finding more info, remove that radeon. leave your 750ti's in slots 1 and 2 then configure in bios primary as the gpu in second slot does it work? if no: 1) latest bios installed? 2) move your radeon in slot 1, and ghave the 750 in 2 and 3 - any good outcome?

Forgive my ignorance.. what ROM are you referring to? I've taken a look at the GTX580 page (https://www.techpowerup.com/gpudb/270/geforce-gtx-580) and haven't found anything like that. For reference, I'm using the Zotac GTX580 (https://www.newegg.ca/Product/Product.aspx?Item=N82E16814500184) Your card has this part number, as i see from the NE link: ZT-50101-10P (assuming your link and information preswnted by NE is corect) On techpowerup, go to db, vga bios collection. Filter based on your vendor, model. I see 12 bioses for 580 zotac. And also some of them are for same model like yours but probably various firmware versions - the dates appear different. One of them could be the solution to your problem. Probably the most recent? But first, check gridrunner's video tutorial on how to attempt to dump the rom yourself. If you manage to do it, you will be better off with that. As i said, my card was very picky and procedure didn't work... -d

Bonjour Do you need the prompt? You can try to run headless and pass the slot 1 gpu as well. Might work for your card, my 550ti didn't like in first slot to passthrough even after rom added to the vm xml. Secondly, maybe you can tell what card is in slot 1? Probably if very old, the bios considers it "legacy" and will disregard your bios selection...

Faced similar issue with my 550ti. First you should use the rom for the gpu in the xml. I tried to dump it based on the written/video guides. Failed, because of the card probably being too old. So had to use the rom from techpowerup - make sure you get the right part number... 1060 was fine to dumo the bios, so issue due to 550ti Secondly, i assume it's in second slot, from the id. Keep it there, mine didnt worked in first slot, even after using the rom file. Third, even with rom tag in xml, my card is not displaying text mode, but only graphics mode. So what i do now is make all installations and config of the vm with 1060, then change vm settings to 550ti once all os working properly. The vm does not display a thing till the login screen, where it switches to graphic mode Due to this, most os' cannot be installed directly with 550ti. I think only ubuntu gnome 16.04 was ok to install directly with 550ti. But i tried so many, that i'm not 100% sure... Bottom line - old card, more tinkering, more trouble... Hope this helps you overcome your issue -d

You should try to create a second vdisk that is in a common place, lets say in the cache, and mount it in both vm's (define it while editing the vm and add extra disk via the + sign). Maybe unraid allows this kind of definition in both vm's. But probably it cannot be mounted in write mode at same time. Probably second machine that tries to mount will do it in read mode, or fail to mount, or fail to start. Worth to try it -d