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


Recommended Posts

Quick Question/Possible Bug?

 

I replaced the Compile CPU Count field with 1,2,3,5,6,7 from "all" and the build failed and threw a bunch of errors. Using "all" yields a positive result.

 

Am I entering the cores wrong or is there something else at play? I noticed in the log that the "1" was missing and it looked like this: " ,2,3,5,6,7"

Link to comment
Quick Question/Possible Bug?
 
I replaced the Compile CPU Count field with 1,2,3,5,6,7 from "all" and the build failed and threw a bunch of errors. Using "all" yields a positive result.
 
Am I entering the cores wrong or is there something else at play? I noticed in the log that the "1" was missing and it looked like this: " ,2,3,5,6,7"
Try enter like 6 to use 6 core.

Envoyé de mon HD1913 en utilisant Tapatalk

  • Thanks 1
Link to comment
17 minutes ago, MowMdown said:

Quick Question/Possible Bug?

 

I replaced the Compile CPU Count field with 1,2,3,5,6,7 from "all" and the build failed and threw a bunch of errors. Using "all" yields a positive result.

 

Am I entering the cores wrong or is there something else at play? I noticed in the log that the "1" was missing and it looked like this: " ,2,3,5,6,7"

If you want to use only want to use six cores enter as @rachid596 said '6' and not the individual cores if you also want to specify individual cores you have to turn on the advanced options on the upper right corner and select the cores you want to use and at Compile CPU Count you have to enter 6 if you choose 6.

 

Hope that answers the question.

 

CPU Count have to ba a single digit (depending on your core count - could be also double digit if you got for example 16 or even tripple for 128 - but that's very unlikely :D if you have that many cores you are my hero 😃 ).

Link to comment
Quote

If you want to use only want to use six cores enter as @rachid596 said '6' and not the individual cores

 

Got it. I assumed the variable followed the manual method of pinning cores to a docker. The documentation doesn't give an example for a specific number of cores except for "all", which is why I did what I did.

 

Quote

if you also want to specify individual cores you have to turn on the advanced options on the upper right corner and select the cores you want to use

 

I totally forgot I could do it this way... apparently I like doing things the difficult way. 

 

Needless to say I did get it working and it booted up 1st try. Woohoo!

  • Like 1
Link to comment
2 hours ago, MowMdown said:

Got it. I assumed the variable followed the manual method of pinning cores to a docker. The documentation doesn't give an example for a specific number of cores except for "all", which is why I did what I did.

Why did you even want to set it not to compile on all cores?

I recommend turning of everything except this container when building the images to avoid conflicts.

 

 

Link to comment

I am quite new to this but when I try to run the Kernel Helpbuilder container, I receive this error 'Download of Kernel v5.8.13 failed, putting container into sleep mode!' . I've also tried the current build 6.8.3 and receive the same error with Kernel v4.19.107. Also, the container does not go into sleep mode when I receive this output 'Download of Kernel v5.8.13 failed, putting container into sleep mode!' in terminal. 

2020-10-12_12.20.15.log

Link to comment
35 minutes ago, Jase said:

I am quite new to this but when I try to run the Kernel Helpbuilder container, I receive this error 'Download of Kernel v5.8.13 failed, putting container into sleep mode!' . I've also tried the current build 6.8.3 and receive the same error with Kernel v4.19.107. Also, the container does not go into sleep mode when I receive this output 'Download of Kernel v5.8.13 failed, putting container into sleep mode!' in terminal. 

2020-10-12_12.20.15.log 8.5 kB · 0 downloads

Please try if you can open this link Click (this is the Kernel v4.19.107 itself that the container tries to download and should theoreticaly work, I also tried this here and it works just normal).

Is it possible that you stopped the container once while it wasn't finished (that could also lead to that error)?

 

What do you mean that the container doesn't go to sleep (sleep is not the same thing as stopping) - sleep means that the container is running and wait for you to fix the issue you have to restart it manually or stop and start it again.

 

You can place the file that the container is trying to download (that I linked above) in your main folder for the container since it's searching for that file if it's already downloaded.

 

Try this steps:

  • stop the container
  • delete all files from the folder for this container in your appdata directory
  • start the container and wait if it's working

Please let me know if it worked.

 

EDIT: What kind of build are you trying to build?

Link to comment
25 minutes ago, ich777 said:

Please try if you can open this link Click (this is the Kernel v4.19.107 itself that the container tries to download and should theoreticaly work, I also tried this here and it works just normal).

Is it possible that you stopped the container once while it wasn't finished (that could also lead to that error)?

 

What do you mean that the container doesn't go to sleep (sleep is not the same thing as stopping) - sleep means that the container is running and wait for you to fix the issue you have to restart it manually or stop and start it again.

 

You can place the file that the container is trying to download (that I linked above) in your main folder for the container since it's searching for that file if it's already downloaded.

 

Try this steps:

  • stop the container
  • delete all files from the folder for this container in your appdata directory
  • start the container and wait if it's working

Please let me know if it worked.

 

EDIT: What kind of build are you trying to build?

Thank you for your response.

 

No, I didn't stop the container throughout the process. I have tried the link above out and the browser says "This site can’t be reached mirrors.edge.kernel.org took too long to respond." I have now tried on another network and the link does work and downloads the file via a browser. Once I deleted all the contents of the 'kernel' folder I then restarted the container, the process starts again but gets stuck at 'downloading Kernel v4.19.107' ...so I then copied the previous manually downloaded kernel file into the 'kernel' folder but the manager fails again and heads back into sleep mode. 

 

I misunderstood the container going to sleep, I thought it meant that the container would stop automatically. 

 

I'm trying to build the current version 6.8.3. but I would also like to try the latest beta. 

 

---Decompressing bzroot, this can take some time, please wait!---
6158+0 records in
6158+0 records out
3152896 bytes (3.2 MB, 3.0 MiB) copied, 0.0232947 s, 135 MB/s
247253+1 records in
247253+1 records out
126593656 bytes (127 MB, 121 MiB) copied, 12.1978 s, 10.4 MB/s
1207870 blocks
Parallel unsquashfs: Using 32 processors
703 inodes (721 blocks) to write

[===============================================================|] 721/721 100%

created 701 files
created 205 directories
created 2 symlinks
created 0 devices
created 0 fifos
Parallel unsquashfs: Using 32 processors
132 inodes (303 blocks) to write

[===============================================================|] 303/303 100%

created 132 files
created 13 directories
created 0 symlinks
created 0 devices
created 0 fifos
---Downloading Kernel v4.19.107---
---Download of Kernel v4.19.107 failed, putting container into sleep mode!---

Link to comment
23 minutes ago, Jase said:

No, I didn't stop the container throughout the process. I have tried the link above out and the browser says "This site can’t be reached mirrors.edge.kernel.org took too long to respond."

Yes that's why the container isn't working because it can't connect to that site, have you got a instance of PiHole running on your server or something else?

Is the site https://www.kernel.org/ working or is it the same?

 

24 minutes ago, Jase said:

I misunderstood the container going to sleep, I thought it meant that the container would stop automatically. 

Yes, if it would stop then the message would say that ;)

This is because done intentionally because the container only stops if the build is finished (but that doesn't mean that everything compiled or installed correctly because I simply can't monitor each build process or the Nvidia driver installation that easy).

 

26 minutes ago, Jase said:

I'm trying to build the current version 6.8.3. but I would also like to try the latest beta. 

Prebuilt images are on the bottom of the first post if you can't build it yourself because something is stopping you form connecting to kernel.org

 

If you can get that file for 6.8.3 'linux-4.19.107.tar.gz' from somewhere else (but that's nothing I recommend) and place it in the main directory it would also built the container.

 

Eventually you can try it from your mobile phone without being connected through your wifi.

 

The file is reachable and downloadable (attached a screenshot):

grafik.png.155308edcb09d848352ffbbfcf001bcd.png

Link to comment
39 minutes ago, ich777 said:

Yes that's why the container isn't working because it can't connect to that site, have you got a instance of PiHole running on your server or something else?

Is the site https://www.kernel.org/ working or is it the same?

 

Yes, if it would stop then the message would say that ;)

This is because done intentionally because the container only stops if the build is finished (but that doesn't mean that everything compiled or installed correctly because I simply can't monitor each build process or the Nvidia driver installation that easy).

 

Prebuilt images are on the bottom of the first post if you can't build it yourself because something is stopping you form connecting to kernel.org

 

If you can get that file for 6.8.3 'linux-4.19.107.tar.gz' from somewhere else (but that's nothing I recommend) and place it in the main directory it would also built the container.

 

Eventually you can try it from your mobile phone without being connected through your wifi.

 

The file is reachable and downloadable (attached a screenshot):

grafik.png.155308edcb09d848352ffbbfcf001bcd.png

Thank you so much for your help! Danke!

 

So, yes I'm being blocked somewhere and I'm guessing my Unifi firewall is preventing it. Manually moving the downloaded kernel 4.19.107 on another network worked after restarting the container. Now it is stuck with ZFS vmaster. Kernel.org works on the network but just won't download with Unraid. Where would I find ZFS vmaster?

 

---Copying Kernel Image to output folder---
---Downloading ZFS vmaster, please wait!---
---Downloading ZFS vmaster, please wait!---
---Download of ZFS vmaster failed, putting container into sleep mode!---

 

I will download your created images and will try that shortly. 

 

Thanks again.

Edited by Jase
Link to comment
7 minutes ago, Jase said:

Thank you so much for your help! Danke!

 

So, yes I'm being blocked somewhere and I'm guessing my Unifi firewall is preventing it. Manually moving the downloaded kernel 4.19.107 on another network worked after restarting the container. Now it is stuck with ZFS vmaster. Kernel.org works on the network but just won't download with Unraid. Where would I find ZFS vmaster?

 

---Copying Kernel Image to output folder---
---Downloading ZFS vmaster, please wait!---
---Downloading ZFS vmaster, please wait!---
---Download of ZFS vmaster failed, putting container into sleep mode!---

 

I will download your created images and will try that shortly. 

 

Thanks again.

Nice that it works now. Bitteschön. ;)

 

Oh no there is a bug on v6.8.3, please change the ZFS version entry from 'master' to '0.8.5' without quotes 0.8.5 is the latest version and I recommend building with this version (I only changed the v6.9.0 container to build from the master branch on Github since there was no easy way of building a version with ZFS for newer Kernels but now with 0.8.5 the building on newer Kernels is possible).

 

Thanks for reporting that!

 

I will update the template in the CA App but it will take a few hours to update.

Link to comment
4 hours ago, ich777 said:

Nice that it works now. Bitteschön. ;)

 

Oh no there is a bug on v6.8.3, please change the ZFS version entry from 'master' to '0.8.5' without quotes 0.8.5 is the latest version and I recommend building with this version (I only changed the v6.9.0 container to build from the master branch on Github since there was no easy way of building a version with ZFS for newer Kernels but now with 0.8.5 the building on newer Kernels is possible).

 

Thanks for reporting that!

 

I will update the template in the CA App but it will take a few hours to update.

Thanks to your absolute brilliance, I think I managed to get through building my own custom image! 

Herzlichen Dank! 

 

I think what got me was my firewall, which BTW I still don't completely understand, as I turned it off.

But, I'm on a Ubiquiti gateway which has been having issues lately which only added to the confusion and frustration I think.

 

So, here is my enclosed log file. Does it look ok?

 

Thanks again Ich777. 

 

 

2020-10-12_20.10.30.log

Edited by Jase
Link to comment
3 hours ago, Jase said:

Thanks to your absolute brilliance, I think I managed to get through building my own custom image! 

Herzlichen Dank! 

 

I think what got me was my firewall, which BTW I still don't completely understand, as I turned it off.

But, I'm on a Ubiquiti gateway which has been having issues lately which only added to the confusion and frustration I think.

 

So, here is my enclosed log file. Does it look ok?

 

Thanks again Ich777. 

 

 

2020-10-12_20.10.30.log 1.07 MB · 0 downloads

The log doesn't look too good...

On which unraid version are you now? You should always be on that version for that you are building for.

 

Something doesn't seem quite right.

Have you already installed the linuxserver.io build for Nvidia or was something using the GPU when you where building the custom images?

 

Also the patches aren't applied right.

 

How big is the image now? With iSCSI ZFS and Nvidia the bzroot should be about 260MB

Link to comment
On 10/12/2020 at 11:49 PM, ich777 said:

The log doesn't look too good...

On which unraid version are you now? You should always be on that version for that you are building for.

 

Something doesn't seem quite right.

Have you already installed the linuxserver.io build for Nvidia or was something using the GPU when you where building the custom images?

 

Also the patches aren't applied right.

 

How big is the image now? With iSCSI ZFS and Nvidia the bzroot should be about 260MB

Sorry ich777, I haven't had a chance to reply to you sooner. 

 

I'm running Unraid version 6.8.3 and I'm running the Unraid Nvidia image from here...

Nothing was running at the time of the creation of Kernel helper but, I've just realized that maybe I'm not suppose to try and create the new Kernel with my running Unraid Nvidia operating system? If so, that is a little embarrassing. From what I understood at the beginning of the thread, "nVidia Driver installation: If you build the images with the nVidia drivers please make sure that no other process is using the graphics card otherwise the installation will fail and no nVidia drivers will be installed."

 

I thought it was ok to run with Unraid Nvidia image. 😳

Edited by Jase
Link to comment
32 minutes ago, Jase said:

I'm running Unraid version 6.8.3 and I'm running the Unraid Nvidia image from here...

No problem but as I said the prebuilt ones in the first post on the bottom are not very different from the one you run now (except for the newer drivers).

 

As I said if you tell me which build you need and with features I'm here to build you a custom image. ;)

Btw it's really strange that it doesn't work on your machine...

 

33 minutes ago, Jase said:

Unraid Nvidia operating system?

No you can do that but no process (Plex, Emby,...) basically anything that uses the graphics card should run since it prevents the nVidia driver from installing correctly.

That's why I put so much warnings in the Log of turning off all other Containers and VM's when you are building the images.

 

35 minutes ago, Jase said:

"nVidia Driver installation: If you build the images with the nVidia drivers please make sure that no other process is using the graphics card otherwise the installation will fail and no nVidia drivers will be installed."

Yes, no process or VM should use the card.

35 minutes ago, Jase said:

I thought it was ok to run with Unraid Nvidia image. 😳

Don't get this scentece exactly, you can build images with my container when you already installed the driver but no PROCESS or VM should use the card or any card itself.

 

 

Don't get it why it doesn't work on your machine...

 

As I said above you can also use a prebuilt image from the bottom of the first post or contact me and I will build you the images that you need if it doesn't work for you.

Link to comment

What's the correct way to switch over from NVIDIA plugin to this one? Need support for Unraid 6.8.3 with latest NVIDIA drivers.

 

- Do I need to uninstall the NVIDIA plugin first?

- If yes, how to get the GPU UUID?

- Are there additional differences between 6.8.3 with NVIDIA plugin and this one here?

 

Any hints are highly appreciated.

 

Edited by hawihoney
Link to comment
17 minutes ago, hawihoney said:

Do I need to uninstall the NVIDIA plugin first?

No, but you can do this since my approach to reading out the UUID and some other things is a little different.

 

17 minutes ago, hawihoney said:

If yes, how to get the GPU UUID?

You can get it with the linuxserver.io plugin or you install the Unraid-Kernel-Helper Plugin

 

17 minutes ago, hawihoney said:

Are there additional differences between 6.8.3 with NVIDIA plugin and this one here?

Not really, it's basically the same.

 

17 minutes ago, hawihoney said:

Any hints are highly appreciated.

Read the first post very carefully. If you are building it on your own then make sure to turn off every container that can use the graphics card (if some process is using the card the installation of the driver will fail at build time).

 

You have to manually place the files that beeing created on your USB boot drive (always backup the existing files to your local computer in case something goes wrong).

A full nVidia build with the latest driver is about 250MB in size in terms of the 'bzroot' file.

Wait for the container to stop then the build process is over, keep an eye on the logs.

For further questions I am here but on weekends not really all the time... :P

 

You can also download a prebuilt one from the bottom of the first post.

Link to comment
22 hours ago, ich777 said:

You can get it with the linuxserver.io plugin or you install the Unraid-Kernel-Helper Plugin

I do have the GPU UUID in my Plex docker already. I took this UUID from the LSIO Nvidia plugin that I'm using currently. I asked my question because you mentioned in this thread somewhere, that the GPU UUID might change, when switching over to your approach. Just curious. 

 

Edit: What about Device IDs like "IOMMU group 36:[1000:0097]"? Can they change too?

 

Edited by hawihoney
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.