[Support] ich777 - AMD Vendor Reset, CoralTPU, hpsahba,...


Recommended Posts

1 hour ago, ich777 said:

As @SimonF wrote above install the iSCSI Plugin and follow the steps that are displayed.

 

I use it on a daily basis since beta35 with my main system and never had a problem.

 

Seems to be working, and was pretty darn simple (as said, just install the plugin and configure).  I'm sure people who want it will figure it out.  ;)

  • Like 1
Link to comment
1 minute ago, Overtaxed said:

 

Seems to be working, and was pretty darn simple (as said, just install the plugin and configure).  I'm sure people who want it will figure it out.  ;)

Eventually Limetech will create a GUI that integrates nicely into Unraid itself.

 

As you can see now the plugin is not very polished and well looks not that good because I'm not really into HTML, PHP,... (but at least it is usable I think and the basic functionality is in there)... :D

Link to comment
35 minutes ago, ich777 said:

Eventually Limetech will create a GUI that integrates nicely into Unraid itself.

 

As you can see now the plugin is not very polished and well looks not that good because I'm not really into HTML, PHP,... (but at least it is usable I think and the basic functionality is in there)... :D

Honestly, I think your plug-in looks great!  The thing I found confusing is what needs to be built in/added to get it working (because this thread goes back to a time before it was in the kernel).  It was really dead simple, install 6.9, install plug in, configure, done.  

 

That said, I used to admin a large (Netapp) based iSCSI environment, so.. I'm pretty familiar with how all the pieces fit together, target/initiator, LUN, mapping, etc.  Certainly would have been harder if I didn't have that experience.  

 

I'm running the iSCSI LUN to VMware, and, when I tried this with an RC, I had some issues with disconnects.  File based LUN on my cache pool (2 drives) would get random disconnects/freeze ups.  I'll see it I have the same issue this time, but, so far, it's rock solid and fast.

  • Like 2
Link to comment
On 2/27/2021 at 2:04 PM, ich777 said:

This is a longer story as you might think with the other user... :D

If this user wasn't that offensive the 6.8.3 Kernel-Helper would still be around (only for your information, I pulled the Kernel-Helper out of the repo entirely and thought a very long time about releasing it again and in which form).

I get Your point also. I am also developing some things for fun, but never have time to publish all :/ recently I am at least trying to do some photo/screenshot story if I get something cool done. I can see that Your kernel-build docker works or partially works for unraid 6.8.3.

EDIT: Ok to be precise script downloads source codes, part of kernel has compiled, however at some point gcc has crashed. I had to login to root at unraid and killall gcc, killall cc1 killall as. Probably some extra effort can be done to make compilation fully sucessfull. I think this error is about some extra patches being applied. For me it is important to compile battery and ac_supply modules, so I will not investigate any further. I am planning to add these two modules manually by using modprobe, without touching original kernel. My unraid's uptime is 224 day, so I won't mess with such a stable kernel.

 

I set variable UNRAID_V = 6.8.3 in docker configuration and CUSTOM_MODE=true.

Then in docker console:

apt-get update

apt-get install mc

then I can use: mc

 

So I have full comfort of editing Your build script, which is cool I must admit. Some good work done here :) Thanks and don't get dicouraged.

 

p.s.

I do not mount any folders as it is shown on dockerhub. I preffer to copy new kernel manually. It is simple and safe to be done :)

 

Edited by Kacper
  • Like 1
Link to comment
10 hours ago, Kacper said:

I get Your point also. I am also developing some things for fun, but never have time to publish all :/ recently I am at least trying to do some photo/screenshot story if I get something cool done.

Only to attack me I think... :D

Joking aside, he needed it for 6.8.2 but for special reasons I don't support "outdated" version as said above I also never released a version for 6.8.2 and also it had a bug in there...

 

10 hours ago, Kacper said:

So I have full comfort of editing Your build script, which is cool I must admit. Some good work done here :)

Really much appreciated... :)

 

10 hours ago, Kacper said:

Thanks and don't get dicouraged.

Of course not, it hurts a little while but I recover fast from such "attacks"... :D

 

10 hours ago, Kacper said:

I do not mount any folders as it is shown on dockerhub. I preffer to copy new kernel manually. It is simple and safe to be done :)

You don't have to mount the folders because the /boot folder is only for backup creation and the other "main" folder is super cool for me because I actually build all my Plugins in it so I don't have to redownload all the drivers again and again and again... :)

Link to comment

I was just about to give up on getting my RX 580 to not take down my whole server all the time when I saw this in my system log:

 

root@server:~# dmesg | grep vfio
[  105.052090] vfio-pci 0000:01:00.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=none
[  107.282570] vfio-pci 0000:01:00.0: enabling device (0002 -> 0003)
[  107.282661] vfio-pci 0000:01:00.0: AMD_POLARIS10: version 1.1
[  107.282662] vfio-pci 0000:01:00.0: AMD_POLARIS10: performing pre-reset
[  107.295589] vfio-pci 0000:01:00.0: AMD_POLARIS10: performing reset
[  107.295594] vfio-pci 0000:01:00.0: AMD_POLARIS10: CLOCK_CNTL: 0x0, PC: 0x2b4c
[  107.295595] vfio-pci 0000:01:00.0: AMD_POLARIS10: performing post-reset
[  107.321575] vfio-pci 0000:01:00.0: AMD_POLARIS10: reset result = 0

 

You are amazing ich777!

 

The only problem I'm having now is that VMs seem to crash during launch with the GPU selected 😅 funny how that seems like a smaller problem. I can see it says that there are no available reset mechanisms. I've attached the logs from a fresh Windows 10 VM. Do you guys have any idea what's going on?

 

 

doonserver-diagnostics-20210306-1948.zip

Windows-10-logs.txt

Edited by lnxd
moved some logs into a .txt file
  • Like 1
Link to comment
51 minutes ago, lnxd said:

The only problem I'm having now is that VMs seem to crash during launch with the GPU selected 😅

So it is working now or not...? :D

 

Have you bound the card to VFIO?

Eventually try to boot with Legacy if you are booting with UEFI into Unraid and also try to select SeaBIOS instead of OVMF if you are having OVMF selected.

  • Thanks 1
Link to comment

This is for people running unraid on LAPTOP!

 

Thanks to this kernel plugin I have managed to compile ac.ko.xz and battery.ko.xz modules. As soon as I have done this, I have noticed that new version of unraid was released :P funny thing. Later I am planning to compile this two modules for new unraid. For now I am uploading version 6.8.3, as I will wait for a while with updating my server.

 

Advantage of this procedure is that oryginal unraid kernel stays untouched. To install it is required to copy module files somewhere on hard drive, for example /boot/myCustomModules.

Then enter following command:

insmod ac.ko.xz

ismod battery.ko.xz

 

Battery information and ac adapter info will be avaliable under folder:

/sys/class/power_supply/

 

Now it is required to write some user script that will shutdown unraid when battery is bellow 30% and ac_power = 0 :)

myCustomModules6_8_3.zip

  • Like 1
Link to comment
1 hour ago, ich777 said:

So it is working now or not...? :D

Booting in Legacy mode fixed that problem as well. Thank you so much! You saved me giving up on using GPUs in my Unraid server! I was ready to go out and buy a second PC. 

 

I wonder if there is something wrong with my GPU that makes it crash so often 🤔 but at least now I can start to investigate the underlying problem without putting my data at risk and bothering my family. Thank you so much!!!!!!!!! 

 

Edit: 

I can't believe it, I've started and stopped VMs 5-6 times while setting up an underclock, the GPU has even crashed a linux VM, and I haven't had to restart the whole server once. Wooh!

Edited by lnxd
  • Like 1
Link to comment
45 minutes ago, lnxd said:

I can't believe it, I've started and stopped VMs 5-6 times while setting up an underclock, the GPU has even crashed a linux VM, and I haven't had to restart the whole server once. Wooh!

Yes, with the gnif/vendor-reset patch, a reboot of the Server shouldn't be necessary because the reset patch makes it possible so that the card actually resets... :D

  • Like 1
Link to comment
1 hour ago, Kacper said:

This is for people running unraid on LAPTOP!

 

Thanks to this kernel plugin I have managed to compile ac.ko.xz and battery.ko.xz modules. As soon as I have done this, I have noticed that new version of unraid was released :P funny thing. Later I am planning to compile this two modules for new unraid. For now I am uploading version 6.8.3, as I will wait for a while with updating my server.

 

Advantage of this procedure is that oryginal unraid kernel stays untouched. To install it is required to copy module files somewhere on hard drive, for example /boot/myCustomModules.

Then enter following command:

insmod ac.ko.xz

ismod battery.ko.xz

 

Battery information and ac adapter info will be avaliable under folder:

/sys/class/power_supply/

 

Now it is required to write some user script that will shutdown unraid when battery is bellow 30% and ac_power = 0 :)

myCustomModules6_8_3.zip 9.76 kB · 1 download

Should I create a Plugin and integrate the script so that it watches if the battery goes down to 30% so you actually only need to download the Plugin and everything is handeled by it?

 

It would be possible... :)

Link to comment

Hi,

I have created mini howto with script that actually does the job for me. Look here:

 

I am sure it can be done better way, e.g. as a plugin.

I have read that in kernel there are is other battery module (sbs), that will work with other, maybe newer laptops. My solution uses ac and battery modules and works with thinkpads and probably many more laptops.

Making universal plugin might require community testing on different machines or reading docs for sbs kernel power module.

 

Anyway, I have accomplished what I wanted and shared my results :) Thanks for help.

I have added kernel modules for unraid 6.9.0. In this post I will attach sbs module, if anyone has pc that is using sbs then maybe can share some knowledge.

 

Best regards,

Kacper

 

myCustomModules6_9_0.zip

Edited by Kacper
Added kernel modules for unraid 6.9.0
  • Like 1
Link to comment

Been using this for some time - love it and thanks!

Just upgraded to 6.9.0 - very smooth. Was worried but just use the docker and plug in and you will be fine.

Question: Is there any testing or anything else we should do for 6.9.0? Again, all looks good. I only use my GPU for FAH, Plex and Unmanic...

Thanks again for the work here!!

  • Like 1
Link to comment
52 minutes ago, TexasDave said:

Been using this for some time - love it and thanks!

Appreciated! :)

 

You don't actually have to use the Unraid-Kernel-Helper if you only need Nvidia, DVB or iSCSI since I've already created Plugins that you can install from the CA App and they will install the Drivers or Runitmes.

 

You need to use it for example if you want to integrate the gnif/vendor-reset patch, that's currently only possible with the Unraid-Kernel-Helper or if you are like me and like images with everything integrated.

 

54 minutes ago, TexasDave said:

Question: Is there any testing or anything else we should do for 6.9.0? Again, all looks good. I only use my GPU for FAH, Plex and Unmanic...

No, the Container is already updated and should work just fine. :)

 

 

Link to comment

Just updated from 6.9.0 - Beta35 (everything worked). So I updated to 6.9.0 and it booted correctly. Just needed to update the Nvidia driver, so I updated it and had to reboot. All went well except that not unraid on the server itself will not display the GUI, I just get a flashing cursor. I can access the GUI from another pc as usual, but I can no longer get the GUI on the server to display.

 

It worked fine on 6.9.0 Beta35 using this plugin. but with the update seems something broke. not sure what.

 

Thoughts?

Link to comment
3 hours ago, almulder said:

Nvidia driver

Why don't you use the Nvidia-Driver Plugin from the CA App instead of this container?

 

3 hours ago, almulder said:

All went well except that not unraid on the server itself will not display the GUI, I just get a flashing cursor.

This is really strange, do you get a output when you turn on your server (BIOS initialization), when does the display disappear or better speaking when did the flashing cursor show up? Have you connected the Display cable to the onboard graphics or the Nvidia Card, what is set in the BIOS, are you booting with UEFI or Legacy?

Link to comment
10 minutes ago, ich777 said:

Why don't you use the Nvidia-Driver Plugin from the CA App instead of this container?

 

This is really strange, do you get a output when you turn on your server (BIOS initialization), when does the display disappear or better speaking when did the flashing cursor show up? Have you connected the Display cable to the onboard graphics or the Nvidia Card, what is set in the BIOS, are you booting with UEFI or Legacy?

I do use the Nvidia-Drive Plugin, not the container. Been using it since you made made it. :)

 

So I was on 6.9 beta 35 with your Plugin and all worked fine, booted to GUI without issue.  Now I updated to 6.9.0 Stable and upon reboot it went into the GUI as expected, then I re installed your plugin, rebooted and now no GUI, but everything else works, its just no local GUI. I get GUI from my web browser, plex transcodes as before, VM pass thought still

works as expected.

Edited by almulder
Link to comment
3 minutes ago, almulder said:

I do use the Nvidia-Drive Plugin, not the container. Been using it since you made made it. :)

Oh sorry... but then this is the wrong thread... :D

Anyways, please try to boot with Legacy if you are booting with UEFI...

 

EDIT: Nothing changed, except that I switched the driver over to my Github and I build it for now because I got many requests about newer drivers (RTX3xxx series).

Link to comment
Just now, almulder said:

A my bad LOL. (And yes Legacy boot) I will repost in other thread once I find it.

 

Just click on Support Thread on the Plugin page within Unraid itself. ;)

 

But it has nothing changed, I can only think of a driver issue, can you try to boot not in GUI mode if you see an output?

  • Like 1
Link to comment
6 minutes ago, AnnabellaRenee87 said:

I'm sorry to ask a dumb question, but Is the Unraid-Kernel-Helper not on CA right now? I'm still on 6.9 RC2.

Because I marked it now that you have to be at least on 6.9.0 stable.

 

Also if you want to build it for 6.9.0 or 6.9.1 you should be on that version and then build it, or at least that's my recommendation. :)

Link to comment
Just now, ich777 said:

Because I marked it now that you have to be at least on 6.9.0 stable.

 

Also if you want to build it for 6.9.0 or 6.9.1 you should be on that version and then build it, or at least that's my recommendation. :)

Will I need Internet to do that? I won't have internet if I can't do the RMRR patch for my PFSense VM.

Link to comment

Hi @ich777,

I have been using your wonderfull kernel helper docker now for a while (mainly to include the gnif/vedor-reset patch for my gpu), but i have a funny annoyance with including zfs.
Every time when i upgrade , the first build does not have zfs working. So it boots into the new build os , but the zfs commands are not available. Then as a second run i build again , after cleaning the output dir etc and without changing anything in the docker settings, just rerun. Put that version in place and then after the boot into that second build zfs works.

Is that expacted behavior that zfs can not be active when building with zfs included ? Or do i have to be on the target release to build the target release with options like zfs etc ?
I was bittten today again by it when updating from 6.9.0 to 6.9.1 and my dockers (30+) autostart but on non mounted zfs. So i have to cleanup all the mess they create in the wrong place (just in directories under the non mouted mountpoint) instead of in the zfs where they all have their persistant storage locations .

Link to comment
  • ich777 changed the title to [Support] ich777 - AMD Vendor Reset, CoralTPU, hpsahba,...

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.