Unraid / Docker #43
Comments
|
It's not tested, but most likely this patch can be used with any Linux system, where Nvidia Driver can be installed. Speaking of Docker, you probably will have to patch driver on a host system. So, it worth trying. |
|
Can this help? |
|
Ok, i have PLEX Inc docker working on Unraid 6.7.0-rc3 with Unraid-Nvidia Plugin mentionned above. I also create a script to enable NVDEC on PLEX since it's not yet supported (found on PLEX forum). Now, I would like to patch, but I got this error
|
|
@pducharme It looks like patch.sh successfully made backup before library edit, but failed to apply changes because /usr filesystem is read-only. I'm not familiar with docker, but I think if /usr actually belong to host system (on top which docker is runned), you should run patch on host. If /usr belongs to immutable docker image, you have to edit image or create overlay. In other words, you have to make /usr writable. But there is always another option:
I know this response pretty vague, but I have no unraid/docker instance and can only guess. |
@pducharme did you run this command on a container or on the host? Did you try to do it inside the nvidia-plugin container? I think it should work there. |
|
Just found this news! https://forums.unraid.net/topic/77813-plugin-linuxserverio-unraid-nvidia/ |
|
They does not want us to ask questions on the forum that can trigger nvidia to shutdown their driver plugin, so I guess we have to continue the discussion here :-) |
|
It worked when I ran it on the host. Confirmed with 5 concurrent transcode on my Plex server 😀 |
That's the exact same plugin? Right? :) |
|
@niXta1 sorry got alittle carried away :) |
Nice! Have you set this up in the user scripts plugin? |
|
@rix1337 No, I just ran the patch.sh from the SSH on my Unraid host. Should I use this with User Scripts plugin? I thing it survive reboots, no? |
|
Anyone know of any overviews of how many sessions are actually supported by each card? Wondering if I should use a 980 or a 1060, my box is coloed so don't have too much time to fiddle around with it. |
|
Hi! See this link (also referenced on top page of this repo). 1060 has newer nvenc chip and supports more codec formats |
|
@Snawoot I was thinking of a list over unlocked cards to see how many streams they would handle ^^ But yeah 1060 would probably be a good choice based on the wider support alone. |
@jimlei it’s hard to say unless you use the exact same video files. They handle between ~5-100 streams I guess. |
How did you get it to install? I get
|
|
So there is a new version out of unraid/nvidia and the patch now fails, stating that the driver is not supported. Anyone else seeing this issue? I can't freaking remember where I dropped the patch.sh in on my Unraid system initially. Anyone know where the patch.sh file goes for the Nvidia drivers? I'm currently searching my entire system for "patch.sh" via Krusader docker but it's taking forever and a day.... |
|
@usafle not sure where you put it, but I have it in 'settings -> user scripts' to run on array start. |
|
Thanks, but I'm looking for the actual location on the Unraid OS to put the 'patch.sh' file so I can run the command:
I forgot where the file goes.... sorry if I wasn't clear enough earlier... it was late, I was annoyed LoL |
|
Go to settings -> user scripts, create a new script, paste the content, and manage it from there. |
|
I already have the user script... I need to run the place the patch.sh file in the same directory of the nvidia drivers to patch them.... that's what I'm asking.. where the NVidia drivers are located to patch on an Unraid OS.... Or am I really not doing this correctly at all? |
|
You can run it from anywhere. |
|
Soon as I tried to run it from 'anywhere' on the OS, I got this: ./patch.sh: line 7: syntax error near unexpected token When I did it the first time (and I wasn't paying attention to HOW I was doing it) I placed it next to the Nvidia drivers in Unraid. It worked. |
|
you need to copy the rawfile and that text only. Sounds like you have bad characters in the content |
|
Don't use notepad or notepad++, use Sublime Text 3. The other two will
insert hidden characters that screw up the script. I had the same issue.
…On Wed, Jul 10, 2019 at 9:07 AM Jeffrey Cowen ***@***.***> wrote:
you need to copy the rawfile and that text only. Sounds like you have bad
characters in the content
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#43?email_source=notifications&email_token=ALRGFCOWU2G2D6LSP6NXL33P6XUJHA5CNFSM4GUJCMQ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZTSK2Y#issuecomment-510076267>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ALRGFCMRGB5LXXPGF6Y7TULP6XUJHANCNFSM4GUJCMQQ>
.
|
|
i have 6.7.2 and all is working fine , i use two scrips the first one unlocks the driver and the second one is supposed to have the plex docker to use the transcoder2 so it will use the nvidia card for encode and decode at least thats how i understand it #!/bin/bash
# halt on any error for safety and proper pipe handling
set -euo pipefail ; # <- this semicolon and comment make options apply
# even when script is corrupt by CRLF line terminators (issue #75)
# empty line must follow this comment for immediate fail with CRLF newlines
backup_path="/opt/nvidia/libnvidia-encode-backup"
silent_flag=''
rollback_flag=''
print_usage() { printf '
SYNOPSIS
patch.sh [OPTION]...
DESCRIPTION
The patch for Nvidia drivers to increase encoder sessions
-s Silent mode (No output)
-r Rollback to original (Restore lib from backup)
-h Print this help message
'
}
while getopts 'rsh' flag; do
case "${flag}" in
r) rollback_flag='true' ;;
s) silent_flag='true' ;;
*) print_usage
exit 1 ;;
esac
done
if [[ $silent_flag ]]; then
exec 1> /dev/null
fi
declare -A patch_list=(
["375.39"]='s/\x85\xC0\x89\xC5\x75\x18/\x29\xC0\x89\xC5\x90\x90/g'
["390.77"]='s/\x85\xC0\x89\xC5\x75\x18/\x29\xC0\x89\xC5\x90\x90/g'
["390.87"]='s/\x85\xC0\x89\xC5\x75\x18/\x29\xC0\x89\xC5\x90\x90/g'
["396.24"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["396.26"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["396.37"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g' #added info from https://github.com/keylase/nvidia-patch/issues/6#issuecomment-406895356
# break nvenc.c:236,layout asm,step-mode,step,break *0x00007fff89f9ba45
# libnvidia-encode.so @ 0x15a45; test->sub, jne->nop-nop-nop-nop-nop-nop
["396.54"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.48"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.57"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.73"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.78"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.79"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.93"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["410.104"]='s/\x85\xC0\x89\xC5\x0F\x85\x96\x00\x00\x00/\x29\xC0\x89\xC5\x90\x90\x90\x90\x90\x90/g'
["415.18"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x40\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["415.25"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x40\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["415.27"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x40\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["418.30"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x40\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["418.43"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x40\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["418.56"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x40\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["418.74"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x0f\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["430.09"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x0f\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
["430.14"]='s/\x00\x00\x00\x84\xc0\x0f\x84\x0f\xfd\xff\xff/\x00\x00\x00\x84\xc0\x90\x90\x90\x90\x90\x90/g'
)
declare -A object_list=(
["375.39"]='libnvidia-encode.so'
["390.77"]='libnvidia-encode.so'
["390.87"]='libnvidia-encode.so'
["396.24"]='libnvidia-encode.so'
["396.26"]='libnvidia-encode.so'
["396.37"]='libnvidia-encode.so'
["396.54"]='libnvidia-encode.so'
["410.48"]='libnvidia-encode.so'
["410.57"]='libnvidia-encode.so'
["410.73"]='libnvidia-encode.so'
["410.78"]='libnvidia-encode.so'
["410.79"]='libnvidia-encode.so'
["410.93"]='libnvidia-encode.so'
["410.104"]='libnvidia-encode.so'
["415.18"]='libnvcuvid.so'
["415.25"]='libnvcuvid.so'
["415.27"]='libnvcuvid.so'
["418.30"]='libnvcuvid.so'
["418.43"]='libnvcuvid.so'
["418.56"]='libnvcuvid.so'
["418.74"]='libnvcuvid.so'
["430.09"]='libnvcuvid.so'
["430.14"]='libnvcuvid.so'
)
NVIDIA_SMI="$(which nvidia-smi)"
if ! driver_version=$("$NVIDIA_SMI" --query-gpu=driver_version --format=csv,noheader,nounits | head -n 1) ; then
echo 'Something went wrong. Check nvidia driver'
exit 1;
fi
echo "Detected nvidia driver version: $driver_version"
if [[ ! "${patch_list[$driver_version]+isset}" || ! "${object_list[$driver_version]+isset}" ]]; then
echo "Patch for this ($driver_version) nvidia driver not found." 1>&2
echo "Available patches for: " 1>&2
for drv in "${!patch_list[@]}"; do
echo "$drv" 1>&2
done
exit 1;
fi
patch="${patch_list[$driver_version]}"
object="${object_list[$driver_version]}"
declare -a driver_locations=(
'/usr/lib/x86_64-linux-gnu'
'/usr/lib/x86_64-linux-gnu/nvidia/current/'
'/usr/lib64'
"/usr/lib/nvidia-${driver_version%%.*}"
)
dir_found=''
for driver_dir in "${driver_locations[@]}" ; do
if [[ -e "$driver_dir/$object.$driver_version" ]]; then
dir_found='true'
break
fi
done
[[ "$dir_found" ]] || { echo "ERROR: cannot detect driver directory"; exit 1; }
if [[ $rollback_flag ]]; then
if [[ -f "$backup_path/$object.$driver_version" ]]; then
cp -p "$backup_path/$object.$driver_version" \
"$driver_dir/$object.$driver_version"
echo "Restore from backup $object.$driver_version"
else
echo "Backup not found. Try to patch first."
exit 1;
fi
else
if [[ ! -f "$backup_path/$object.$driver_version" ]]; then
echo "Attention! Backup not found. Copy current $object to backup."
mkdir -p "$backup_path"
cp -p "$driver_dir/$object.$driver_version" \
"$backup_path/$object.$driver_version"
fi
sha1sum "$backup_path/$object.$driver_version"
sed "$patch" "$backup_path/$object.$driver_version" > \
"$driver_dir/$object.$driver_version"
sha1sum "$driver_dir/$object.$driver_version"
ldconfig
echo "Patched!"
fiand #!/bin/bash
############################### DISCLAIMER ################################
# This script now uses someone elses work! #
# Please visit https://github.com/revr3nd/plex-nvdec/ #
# for the author of the new transcode wrapper, and show them your support!#
# Any issues using this script should be reported at: #
# https://gist.github.com/Xaero252/9f81593e4a5e6825c045686d685e2428 #
###########################################################################
# This is the download location for the raw script off github. If the location changes, change it here
plex_nvdec_url="https://raw.githubusercontent.com/revr3nd/plex-nvdec/master/plex-nvdec-patch.sh"
patch_container_path="/usr/lib/plexmediaserver/plex-nvdec-patch.sh"
# This should always return the name of the docker container running plex - assuming a single plex docker on the system.
con="$(docker ps --format "{{.Names}}" | grep -i plex)"
# Uncomment and change the variable below if you wish to edit which codecs are decoded:
#CODECS=("h264" "hevc" "mpeg2video" "mpeg4" "vc1" "vp8" "vp9")
# Turn the CODECS array into a string of arguments for the wrapper script:
if [ "$CODECS" ]; then
codec_arguments=""
for format in "${CODECS[@]}"; do
codec_arguments+=" -c ${format}"
done
fi
echo -n "<b>Applying hardware decode patch... </b><br/>"
# Grab the latest version of the plex-nvdec-patch.sh from github:
echo 'Downloading patch script...'
wget -qO- --show-progress --progress=bar:force:noscroll "${plex_nvdec_url}" | docker exec -i "$con" /bin/sh -c "cat > ${patch_container_path}"
# Make the patch script executable.
docker exec -i "$con" chmod +x "${patch_container_path}"
# Run the script, with arguments for codecs, if present.
if [ "$codec_arguments" ]; then
docker exec -i "$con" /bin/sh -c "${patch_container_path}${codec_arguments}"
else
docker exec -i "$con" /bin/sh -c "${patch_container_path}"
fi |
|
Sorry that didnt copy in how i thought it would |
|
@Rooster237 edited |
|
thanks that looks better . there are two scripts there the first one ends at the AND |
|
If wget -O patch.sh "https://raw.githubusercontent.com/keylase/nvidia-patch/master/patch.sh"otherwise if curl -o patch.sh -s "https://raw.githubusercontent.com/keylase/nvidia-patch/master/patch.sh" |
|
I'm truly sorry all. I was trying to place the actual PATCH.SH file into my UnRaid OS and then from a terminal window run the command I didn't realize that simply deleting the old user script I had set up and then copying pasting the "new" patch.sh contents into a new user script would allow me to accomplish the same thing. Hopefully I'm making sense as to what I was attempting to do incorrectly so someone else reading this will understand that you DO NOT have to do that. I've copied/pasted the contents of the patch.sh into a new user script and all is working... No idea where I got the notion that I had to put the actual patch.sh file "next to" the Nvidia driver on the Unraid OS.... I should really go back and delete all my posts so I don't look that stupid. LoL |
|
Has the plex decoding side of this changed recently? I ask because I installed the patch into my unraid as above, (it works first time thankyou). But I did not install the second patch which above is said to be for decode. Nevertheless when I run nvidia-smi -s u (which I'm told displays the encode and decode activity), it DOES list decode, which would seem to mean the GPU is handling the decoding also. Any thoughts? e.g. gpu sm mem enc decIdx % % % % |
Plex does native nvdec since recently |
|
Great, so for completeness for the next reader, the second script above is no longer needed. |
|
I had created a user script with this following that runs at Start Up of Array. This automatically grabs the latest patch script that has the newest drivers editions included so you don't have to manually copy and paste the latest script. |
|
That's great. So just copy/paste the above into a new user script and that's it? Edit: Guess not because I did that, ran the script and got these errors (that I have no idea what they mean because I'm a NOOB!) |
Nice! good idea, instead of having to reupdate the user script at each new version! |
Yup. That should handle it all automatically so you shouldn't have to worry about it when updating to new Unraid Nvidia. It appears you may have an issue executing the bash command. you could try with this script. wget https://raw.githubusercontent.com/keylase/nvidia-patch/master/patch.sh EDIT: I see my issue was with my originial comment having the bash command on the same line. The bash ./patch.sh should be on a new line. |
Tried that, and got this:
Tried that and got this: Not sure what the initial errors are up top at the start of the script... but the end result looks like it works? |
To simplify & sooth OCD, add --no-hsts
Re: unneeded errors in logs just waste io/stor, Stop the NOISE. |
Hello!
Is this something we can use with UnRaid / Docker ?
The text was updated successfully, but these errors were encountered: