charleslam

what i am trying to do is remove 3x1tb hdd and replace with 3x256gb ssd. this will definitely cause problems with the array. macrium reflect makes images of running partitions. so i installed macrium on all 3 of my VM's and created image files of them. saved on my server. i am rebuilding one just to verify it will work after being shrunk. so far i am at 36%. once this has been verified (or i find that it destroys the image *shrugs*) i will move forward with replacing the hdd's with ssd's. im not sure what unraid will do if i delete all hdd's and try to enable the ssd's. id imagine it would say my array cannot be started due to the missing disks. trying to avoid doing the new config, but it might be the simplest.

im testing it now by creating a new test disk for one of the VM's and seeing if it reproduces the image successfully. ill keep ya guys posted. Macrium uses a rescue disc that uses windows PE. right now it has had my cpu's pegged for that VM for a while now. not sure if it crashed or not. yup it crashed. More than one way to skin a cat. created new Vdisk@256gb. set it as a secondary hdd on my test VM. im going try restore a macrium image to that secondary vdisk, and test booting it from my VM3.

I have this board in an intel chassis for my main server. the onboard raid controller is just a crap LSI (i read its a software emulated raid controller??). get a dedicated raid controller. ive had an intel 1156 server, then to a dell poweredge t610, then onto this board with 2x2630L. by far the most quiet setup ive ever had. one thing to note though, if you get it with a intel chassis like i did, the fans may ramp up to max and stay that way. simply power down the whole thing and remove power from both PSU's. start it up again and it will go back to being quiet (if you set it up that way in the bios). also goes without saying, update the bios to the latest. i noticed weird behavior on it on earlier bios revisions. intel boards are really strange now having to update through their efi command shell, but the scripts work well enough to make it less of a hassle. the intel management application is pretty cool too if you have snmp alerts set up on your network. i have yet to try unraid with it so i dont know how well it plays with that.

I believe he is referring to "crunching numbers" as in computational math, and "folding" like Fold@home for finding the cure for cancer. i may be wrong, but thats my guess. i mean what else are you going to do with 144 cpus?

idk call me crazy but i use gaming on all my TV's. so gaming computers it is. just recycled parts from the latest and greatest of yesteryear.

This is why i really dont like the layout of having the option to assign USB devices without passing through the usb controller. certain devices dont work correctly when it is done. and then youre stuck guessing if something broke or not. I believe other people in here have had the same problems with xbox controllers/receivers. typically the worst is when they try adding multiple xbox controllers/receivers iirc. passing through a controller is a pita as shown, but once you do it, its worth it. makes everything a whole lot easier.

is there an easy way to clear the array information?

i used r9 270x's and didnt need to do anything for gpu passthrough. i'd give it a try and go from there.

understood, i think im going to try to tinker with macrium reflect to get things working. but i will keep people updated if i find a way to do it. worst case scenario i will just start from scratch.

so. my original set up was with 1tb platter HDD's making all my VM's run slow. i have 1 per VM for a total of 3 VM's. so 3x1tb + 1x2tb parity. i recently picked up some 256gb SSD's. moving forward, should i just remove 1 hdd with a SSD at a time and will it let me resync? or do i need to scrap the whole thing? i believe i set the vdisk image size to the total size of the drive (1tb). but as i look at the hdd usage, im noticing the vdisks arent taking up the whole disk. Should i just scrap the whole thing and reconfig from scratch or will replacing 1 disk at a time work?

finally got the last bits tidied up. added a usb slot adapter and rearranged which usb controllers go to what. i only use the asmedia usb controller to boot unraid from as it was incredibly unreliable and would give dma read request errors all day long. i still havent gotten to the bottom of that. people say that its this or its that or try this command. all of which never seem to work. kind of sad though i purchased to usb controller cards and they havent come in the mail yet. one of which is a 6 port usb dual usb controller. one via, one asmedia. and i also purchased a pcie 1x riser ribbon and board with a separate TI usb controller (was told they worked the best). Now the only thing to work on is getting SSD's loaded on this. we shall see how this goes.

i have run into those errors on hard drives multiple multiple times. drive is dying and it will impact performance to the point that you cant boot up. the quicker you switch it out the better.

yeah those really didnt work either.

rob from another post states: " So I guess the best advice is to keep looking for updates to firmware, BIOS, and Linux kernel (the kernel core itself as well as the virtualization support, especially IOMMU, and the included drivers). Some workarounds that are generally suggested, with possibly undesirable side-effects (choose only one!) - - iommu=pt - iommu=pt r8169.use_dac=1 - intel_iommu=pt - intel_iommu=igfx_off - iommu=off - intel_iommu=off - replace problem hardware - disable virtualization support in BIOS " i tried the iommu=pt just by itself, and my intel usb controller started working. but i wonder if the - iommu=pt r8169.use_dac=1 will do a more thorough job.

Oct 9 19:50:10 UNRAID kernel: usb 2-1.4: new high-speed USB device number 7 using ehci-pci Oct 9 19:50:22 UNRAID kernel: usb 2-1.4: reset high-speed USB device number 7 using ehci-pci Oct 9 19:50:22 UNRAID kernel: usb 2-1.4: reset high-speed USB device number 7 using ehci-pci Oct 9 19:50:22 UNRAID kernel: usb 2-1.4: USB disconnect, device number 7 Oct 9 19:50:24 UNRAID kernel: usb 2-1.4: new high-speed USB device number 8 using ehci-pci Oct 9 19:50:24 UNRAID kernel: usb 2-1.4: config 1 interface 0 altsetting 0 bulk endpoint 0x1 has invalid maxpacket 64 Oct 9 19:50:24 UNRAID kernel: usb 2-1.4: config 1 interface 0 altsetting 0 bulk endpoint 0x81 has invalid maxpacket 64 Oct 9 19:51:30 UNRAID kernel: usb 2-1.4: USB disconnect, device number 8 Oct 9 19:51:45 UNRAID kernel: usb 2-1.4: new high-speed USB device number 9 using ehci-pci Oct 9 19:51:55 UNRAID kernel: usb 2-1.4: reset high-speed USB device number 9 using ehci-pci Oct 9 19:51:56 UNRAID kernel: usb 2-1.4: reset high-speed USB device number 9 using ehci-pci Oct 9 19:51:56 UNRAID kernel: usb 2-1.4: USB disconnect, device number 9 this is the part of your syslog that makes me wonder. usb resets and then it detects a new usb device, resets then finds another new usb device. i wonder if your bios settings for usb timeouts can be adjusted?

im running into the issue that when upon first boot, i will have to unplug and plug back into the same controller, usb devices work fine. reboot the VM and no amount of unplugging/plugging fixes.

testing now, because people suck. aaaaand finished testing. adding those lines into the config do not fix the problem with ASMedia usb controllers. maybe im adding the lines to the wrong part of the config? label unRAID OS menu default intel_iommu=on intel_iommu=pt intel_iommu=igfx_off kernel /bzimage but regardless this sucks for people with asmedia crap usb controllers.

im really surprised no one else has chimed in on this.

grr. ok. so people are saying that we need to have: A quick fix consists in replacing the boot kernel parameter ‘intel_iommu=on’ with ‘intel_iommu=pt’ “to set up pass through (PT) mode in context mapping entry”. DMAR is then disabled in the GNU/Linux kernel but KVM still benefits IOMMU and interrupt remapping. and ALSO: intel_iommu=igfx_off to not get the dma read error. havent gotten a chance to try it since im doing a rebuild on my array. Sources: https://centoros.wordpress.com/2015/07/22/iommu-dmar-fault/ https://bbs.archlinux.org/viewtopic.php?id=176398

idk man. i was looking through your logs to see anything obvious. im not seeing anything that pops out to me. but then again im pretty noobish myself on this.

so subscribing to this. I have been digging around myself. finding alot of cards that have master, and slave controllers on them. e.g. https://www.newegg.com/Product/Product.aspx?Item=9SIA1DS4NA7781 i wonder if pass through is possible with those types of configurations?

*update* i have found my stability issues were caused due to a faulty hard drive. it runs stable (albeit super slow on platter HDD's). i have purchased SSD's to assist with this. i tested it with a spare 64gb ssd, and speeds were acceptable. i am having a pesky issue with a asmedia usb controller. keeps dying while the computer is on. sometimes i can switch ports and it works but, i think i need to find a pci-e usb card that has 2 controllers on it that support pass through. suggestions?

while i did have it work intermittently, it gets confusing and is a tad misleading that it will work or it will continue to work. i was second guessing my devices or my motherboard. i mean yeah it worked-ish for keyboards and mice (still had that weird experience with the m600 mouse...idk wtf that was all about). *mileage may vary getting these usb devices working as intended* well it wasnt spelled flat out before from what i could find on these forums for the usb headsets. now it is.

Yeah i found this out WAAAAAAAAAAAAAAY after the fact. I heard that at 11 AM this morning thinking "wow i wish i knew that before i started trying to throw this system together". im kind of frustrated that i did not find this out until after all that.

USB passthrough got the USB headsets to work. went through space invader's method. found this prior to your post SSD. just wish it didnt take me so long to find it. kind of pointless to have an interface with options for things that may/may not work like that. i get it, no one wants to try to figure out why usb headsets dont work consistently in the fashion i set up. but leaving that option up is misleading and confusing. luckily i had space for one usb pci-e card. intel has 1 controller - passthrough worked right out the box asmedia has 2 controller - only 1 ever worked and thats with setting intel_iommu=pt. the 2nd controller shows the same address as the first one (1b21:1042) while showing up on a different bus. so that lead me to believe it really was only 1 physical controller and some other voodoo magic was going on. usb pcie card shows up as a VIA usb controller, but strangely enough did not show up on my bus list. manually added the address and its working. *shrugs* im done trying to figure out consistency with this stuff. regardless thats what happened and what i did to get it working.