Unraid OS version 6.6.0-rc1 available

Updated successfully, still having the AMD reboot bug with a Vega 56 GPU though which is a shame, I thought this may be fixed with kernel 4.16, is there anything I should do with the VM post update to try and fix this?

I have the same mistake. I have to test a new VM created and it comes with the same message!

8 minutes ago, bonienl said:

Intel

I have a feeling it's an AMD releated issue with Unraid 6.60-RC1. In earlier Versions i hadn't had any issues like this. Maybe someone with a Ryzen or Threadripper system reading this can test my xml or can try to reproduce this error. 

 

win7_outlook.xml

 

This looks like another Threadripper/Ryzen related issue i posted in another thread couple days ago. The shown core pairings in unraid didn't really match the real pairing of the cores. I did a couple of test to find out which core is on which die and which ones are the threads only. It looks like the pairings presented to the user by unraid isn't correct. This hasn't changed with the 6.60-RC1 version.

 

 

Need involvement of @eschultz to help out.

Quote

5 minutes ago, bastl said:

I have a feeling it's an AMD releated issue with Unraid 6.60-RC1. In earlier Versions i hadn't had any issues like this. Maybe someone with a Ryzen or Threadripper system reading this can test my xml or can try to reproduce this error. 

 

I also have an Intel system and can not edit the vm

5 minutes ago, bastl said:

I have a feeling it's an AMD releated issue with Unraid 6.60-RC1. In earlier Versions i hadn't had any issues like this. Maybe someone with a Ryzen or Threadripper system reading this can test my xml or can try to reproduce this error. 

I created a new VM using your XML on a Threadripper machine.  Was able to edit and save (without any changes).  Then edit, increase the memory and save again.  No error messages here.

1 minute ago, chatman64 said:

I also have an Intel system and can not edit the vm

When I copy+paste your XML file, I get unknown characters in it. Is it possible for you to attach your XML file to your post instead of the text. Thanks.

Quote

3 hours ago, alturismo said:

upgrade worked here fine it seems, one question i have about igpu

 

i915.alpha_support=1  <<-- still needed to get hardware modprobe i915 working on coffee lake cpu´s ?

or can this be "updaated" to 965 (no idea if possible)

 

I found out that the parameter i915.alpha_support = 1 is no longer needed. The HW transcoding in Plex now works with the coffe lake without this command and the call traces are now gone!

Great work!

10 minutes ago, chatman64 said:

I also have an Intel system and can not edit the vm

When I try to edit your VM (based on the XML in your prior post) I get this message box:

 

 

@chatman64 Are you german same as me? I asked because in your screenshot i see you have configured VNC to use a german layout. Maybe something with the localisation causes that error. I already checked if the VNC layout is the issue, but it wasn't the case. Tested a VM without VNC added to it, same error and chossing us layout causes it. And as mentioned earlier, creating a new vm without editing anything caused that error on the first edit afterwards.  

31 minutes ago, bastl said:

It looks like the pairings presented to the user by unraid isn't correct.

Unraid reads the CPU information from linux. What is the output of this command

cat /sys/devices/system/cpu/*/topology/thread_siblings_list

 

17 minutes ago, bastl said:

And as mentioned earlier, creating a new vm without editing anything caused that error on the first edit afterwards.  

Do you have anything special in your syslinux.cfg file or go file which may affect VMs?

 

The error message indicates it can not parse the PCI device address <bus> <slot> <function> (it fails on the 'function' part)

Edited September 2, 2018 by bonienl

Quote

2 minutes ago, bonienl said:

Unraid reads the CPU information from linux. What is the output of this command

cat /sys/devices/system/cpu/*/topology/thread_siblings_list

 

look at the picture

 

Quote

19 minutes ago, bastl said:

Are you german same as me? I asked because in your screenshot i see you have configured VNC to use a german layout. Maybe something with the localisation causes that error. I already checked if the VNC layout is the issue, but it wasn't the case. Tested a VM without VNC added to it, same error and chossing us layout causes it. And as mentioned earlier, creating a new vm without editing anything caused that error on the first edit afterwards.  

 

yes I am German too. I have already experimented with the language setting but even if I create a new vm with English, I have the same Problem.

 

Quote

33 minutes ago, bonienl said:

When I copy+paste your XML file, I get unknown characters in it. Is it possible for you to attach your XML file to your post instead of the text. Thanks.

 

I modified the XML in the old post. I do not know how I can make a backup of the original file.

8 minutes ago, bonienl said:

Unraid reads the CPU information from linux. What is the output of this command

cat /sys/devices/system/cpu/*/topology/thread_siblings_list

 

The command brings up the following:

 

 

For whatever reason every pair is shown twice.

 

My current syslinux config looks like this

 

 

I removed the following today after i tested disabling the ACS override patch.

 

"pcie_acs_override=downstream,multifunction"

 

IOMMU groups are fine now with the current kernel and my updated BIOS yesterday. With the older BIOS from end of last year i wasn't able to passthrough a GPU or the nvme controller without this line. Now it works without. All the issues with core pairings or the noneditable VMS existed on both configurations.

 

I don't really sure if i ever touched the go file. 

 

 

 

 

10 minutes ago, chatman64 said:

I modified the XML in the old post. I do not know how I can make a backup of the original file.

You can find the xml files in /etc/libvirt/qemu

Quote

25 minutes ago, bonienl said:

Do you have anything special in your syslinux.cfg file or go file which may affect VMs?

 

The error message indicates it can not parse the PCI device address <bus> <slot> <function> (it fails on the 'function' part)

 

here my syslinux.cfg:

default menu.c32
menu title Lime Technology, Inc.
prompt 0
timeout 50
label unRAID OS
  menu default
  kernel /bzimage
  append vfio-pci.ids=1912:0014 initrd=/bzroot
label unRAID OS GUI Mode
  kernel /bzimage
  append initrd=/bzroot,/bzroot-gui
label unRAID OS Safe Mode (no plugins, no GUI)
  kernel /bzimage
  append initrd=/bzroot unraidsafemode
label unRAID OS GUI Safe Mode (no plugins)
  kernel /bzimage
  append initrd=/bzroot,/bzroot-gui unraidsafemode
label Memtest86+
  kernel /memtest

 

append vfio-pci.ids=1912:0014 initrd=/bzroot is my USB3 Card

 

and here my go file:

#!/bin/bash
# Start the Management Utility
modprobe i915
chown -R nobody:users /dev/dri
chmod -R 777 /dev/dri
chmod -R 777 /dev/ttyUSB0
/usr/local/sbin/emhttp &

 

Quote

17 minutes ago, bastl said:

You can find the xml files in /etc/libvirt/qemu

 

the path does not exist

2 minutes ago, chatman64 said:

the path does not exist

For me the mounted libvirt.img is accessible under this path via the webterminal

 

ok over ftp I come to the path. here an xml file

W10 Download Papa.xml

Hi guys, i have some problems editing my vm template via gui. I make changes, then press Apply, and nothing happens. Tried to add virtio iso to vm, tried to swich i440fx from 2.11 to 3.0. Nothing worked, no errors, just nothing happens..

here is XML and diag.

Windows 10.xml

unraid-diagnostics.zip

 

PS: my 6.5.3 xml from backup

6.5.3 Windows 10.xml

PPS: i did some tests. Backed-up 6,6 xml to file, then deleted vm template from gui, recreated vm with my img file. VM does not boot (drops into uefi shell). 

Then i returned back 6.6xml using gui.  6.6 Windows 10.xml  VM booted normaly. Now i can edit XML in gui, but after any edit i cant boot my VM (drops to uefi shell).

 

need help 😃

UPD: if i swich to xml view and edit i440fx-2.1.1 to i440fx-3.0 this works. If i change in gui it doesn`t

UPD2: I found what drops to uefi shell. The reason of wrong .img format. I use qcow2, vm was created with qcow2, and every time i save (update) my template via gui unraid write RAW format in xml config and vm drops to uefi.

Edited September 2, 2018 by vanes

FYI, I've tried this for myself just to see if I would have problems also. I haven't had any issues with it.

 

I've changed memory, machine types back and forth between 2.xx and 3.0, changing USB devices, and CPU allocation. No problems whatsoever. And I didn't lose my custom XML edit for emulator pinning!! 😃

19 hours ago, bonienl said:

With this plugin present, it will overwrite the style sheet files for the white and black themes.

The white and black themes in earlier versions had minor updates/corrections. Too small to notice the overwriting, but version 6.6 has a complete new design which doesn't work at all with the old style sheets.

 

Made a propossal (updated install script) to LT to include the deletion of the dynamix plugin when Unraid 6.6 gets installed.

 

You may want to add a warning (or delete the plugin) in Update Assistent, so people are prepared  in advance.

 

Prior behavior on the Update Assistant was that it checked for a version of dynamix.plg with a file time before Jan 1, 2016 and told you to delete them (since latter versions checked on the OS version prior to installing).  Since there's some bug or something in the version check on some of dynamix.plg, Update Assistant now will always tell you to delete dynamix.plg no matter the time stamp.

A reminder: this thread is for general discussion and not meant to report and solve specific problems.  If you report a problem in this thread, instead of creating a separate topic for it in this board, you may receive support but you should not expect it.   PLEASE open separate topics here.

Updated with no noticeable problems so far

Updated all 3 servers and so far so good. Loving the new look. Thanks. 😊

13 hours ago, bastl said:

For whatever reason every pair is shown twice.

It's a consequence of the way the "*" wildcard in the command is expanded to include all threads. So, for example, thread 0 is shown with its pair, thread 16. Then later thread 16 is shown with its pair, thread 0.