Proxmox LXC - Revision history

Deathbybandaid: /* GPU Passthrough */

2023-04-20T17:03:25Z

GPU Passthrough

← Older revision		Revision as of 17:03, 20 April 2023
Line 96:		Line 96:
	====On Host====		====On Host====

			<nowiki>apt install -y build-essential gcc-multilib dkms</nowiki>

	=====Blacklist Nouveau Drivers=====		=====Blacklist Nouveau Drivers=====

	<nowiki>nano /etc/modprobe.d/blacklist.conf</nowiki>		<nowiki>nano /etc/modprobe.d/blacklist-nouveau.conf</nowiki>

	<nowiki>blacklist nouveau</nowiki>		<nowiki>blacklist nouveau
			options nouveau modeset=0</nowiki>

	<nowiki>update-initramfs -u</nowiki>		<nowiki>update-initramfs -u</nowiki>

	<nowiki>reboot</nowiki>		<nowiki>reboot</nowiki>

	~~<nowiki>apt install -y build-essential</nowiki>~~


Line 115:		Line 115:
	[https://www.nvidia.com/Download/index.aspx]		[https://www.nvidia.com/Download/index.aspx]
	<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>
	<nowiki>./NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>./NVIDIA-Linux-x86_64-460.84.run --dkms -s</nowiki>

	======Apt======		======Apt======

Deathbybandaid: /* On Host */

2023-04-03T15:17:32Z

On Host

← Older revision		Revision as of 15:17, 3 April 2023
Line 97:		Line 97:


	Blacklist Nouveau Drivers		=====Blacklist Nouveau Drivers=====

	<nowiki>nano /etc/modprobe.d/blacklist.conf</nowiki>		<nowiki>nano /etc/modprobe.d/blacklist.conf</nowiki>
Line 110:		Line 110:


	Download Nvidia Driver for your card		=====Download Nvidia Driver for your card=====

	=====Manual=====		======Manual======
	[https://www.nvidia.com/Download/index.aspx]		[https://www.nvidia.com/Download/index.aspx]
	<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>
	<nowiki>./NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>./NVIDIA-Linux-x86_64-460.84.run</nowiki>

	=====Apt=====		======Apt======

	<nowiki>nano /etc/apt/sources.list</nowiki>		<nowiki>nano /etc/apt/sources.list</nowiki>
Line 124:		Line 124:
	<nowiki>non-free</nowiki>		<nowiki>non-free</nowiki>

	Create Kernel rules		=====Create Kernel rules=====
	<nowiki>nano /etc/udev/rules.d/70-nvidia.rules</nowiki>		<nowiki>nano /etc/udev/rules.d/70-nvidia.rules</nowiki>

Line 146:		Line 146:
	</nowiki>		</nowiki>

	Update initramfs		=====Update initramfs=====

	<nowiki>update-initramfs -u</nowiki>		<nowiki>update-initramfs -u</nowiki>

	Do a reboot		=====Do a reboot=====
	<nowiki>reboot</nowiki>		<nowiki>reboot</nowiki>

	Find the GPU device number. This usually is 226:0,128		=====Find the GPU device number.=====
			This usually is 226:0,128
	<nowiki>ls -l /dev/dri</nowiki>		<nowiki>ls -l /dev/dri</nowiki>

Line 175:		Line 176:
	Take note of the numbers in the fifth column above 195, 236 and 226 respectively.		Take note of the numbers in the fifth column above 195, 236 and 226 respectively.

	ADD these lines to /etc/pve/lxc/<container-id>.conf		=====ADD these lines to /etc/pve/lxc/<container-id>.conf=====
	<nowiki>		<nowiki>
	lxc.cgroup2.devices.allow: c 195:0 rw		lxc.cgroup2.devices.allow: c 195:0 rw
Line 195:		Line 196:
	lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file		lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file
	</nowiki>		</nowiki>

			=====Setup Nvidia persistanced service=====

			To avoid that the driver/kernel module is unloaded whenever the GPU is not used, we should run the Nvidia provided persistence service. It’s made available to us after the driver install.

			Copy and extract

			<nowiki>cp /usr/share/doc/NVIDIA_GLX-1.0/samples/nvidia-persistenced-init.tar.bz2 .</nowiki>
			<nowiki>bunzip2 nvidia-persistenced-init.tar.bz2</nowiki>
			<nowiki>tar -xf nvidia-persistenced-init.tar</nowiki>
			Remove old, if any (to avoid masked service)

			<nowiki>rm /etc/systemd/system/nvidia-persistenced.service</nowiki>
			Install

			<nowiki>chmod +x nvidia-persistenced-init/install.sh</nowiki>
			<nowiki>./nvidia-persistenced-init/install.sh</nowiki>
			Check that it’s ok

			<nowiki>systemctl status nvidia-persistenced.service</nowiki>
			<nowiki>rm -rf nvidia-persistenced-init*</nowiki>

	====In Container====		====In Container====

Deathbybandaid: /* On Host */

2023-03-23T15:20:50Z

On Host

← Older revision		Revision as of 15:20, 23 March 2023
Line 109:		Line 109:
	<nowiki>apt install -y build-essential</nowiki>		<nowiki>apt install -y build-essential</nowiki>

	[https://www.nvidia.com/Download/index.aspx ~~Download Nvidia Driver for your card~~]
			Download Nvidia Driver for your card

			=====Manual=====
			[https://www.nvidia.com/Download/index.aspx]
	<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>
	<nowiki>./NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>./NVIDIA-Linux-x86_64-460.84.run</nowiki>

			=====Apt=====

			<nowiki>nano /etc/apt/sources.list</nowiki>

			add to your main repo
			<nowiki>non-free</nowiki>

	Create Kernel rules		Create Kernel rules
	<nowiki>nano /etc/udev/rules.d/70-nvidia.rules</nowiki>		<nowiki>nano /etc/udev/rules.d/70-nvidia.rules</nowiki>

Deathbybandaid: /* Mounting a CIFS share in a container */

2023-01-17T17:46:43Z

Mounting a CIFS share in a container

← Older revision		Revision as of 17:46, 17 January 2023
Line 26:		Line 26:
	<nowiki>nano /etc/autofs/DrivePool.conf</nowiki>		<nowiki>nano /etc/autofs/DrivePool.conf</nowiki>

	~~<nowiki>~~DrivePool -fstype=cifs,rw,credentials=/etc/autofs/DrivePool.creds,file_mode=0777,dir_mode=0777 ://IPADDRESS/DrivePool~~</nowiki>~~		DrivePool -fstype=cifs,vers=3.0,rw,relatime,sec=ntlmssp,cache=strict,credentials=/etc/autofs/DrivePool.creds,uid=1001,forceuid,gid=111,forcegid,file_mode=0777,dir_mode=0777,nounix,serverino,mapposix,rsize=1048576,wsize=1048576,echo_interval=60,actimeo=1 ://IPADDRESS/DrivePool

	<nowiki>nano /etc/autofs/DrivePool.creds</nowiki>		<nowiki>nano /etc/autofs/DrivePool.creds</nowiki>
Line 40:		Line 40:
	<nowiki>nano /etc/auto.master</nowiki>		<nowiki>nano /etc/auto.master</nowiki>

	~~<nowiki>~~/FS1 /etc/autofs/DrivePool.conf --ghost~~</nowiki>~~		/FS1 /etc/autofs/DrivePool.conf --ghost --timeout 0 --verbose
	Restart The Service		Restart The Service
	<nowiki>service autofs restart</nowiki>		<nowiki>service autofs restart</nowiki>

Deathbybandaid: /* Nvidia */

2023-01-13T18:09:00Z

Nvidia

@@ Line 122: / Line 122: @@
   KERNEL=="nvidia_uvm", RUN+="/bin/bash -c '/usr/bin/nvidia-modprobe -c0 -u && /bin/chmod 0666 /dev/nvidia-uvm*'"
   </nowiki>
 Do a reboot
   <nowiki>reboot</nowiki>
 Find the GPU device number. This usually is 226:0,128
   <nowiki>ls -l /dev/dri</nowiki>
@@ Line 172: / Line 189: @@
   <nowiki>./NVIDIA-Linux-x86_64-460.84.run --no-kernel-module</nowiki>
 ===AMD Radeon===
 ====On Host====

Deathbybandaid: /* GPU Passthrough */

2023-01-13T17:31:00Z

GPU Passthrough

← Older revision		Revision as of 17:31, 13 January 2023
Line 90:		Line 90:
	Install pve-headers, before and after you do any driver installs on host a below		Install pve-headers, before and after you do any driver installs on host a below
	<nowiki>apt install pve-headers</nowiki>		<nowiki>apt install pve-headers</nowiki>
			<nowiki>apt install -y pve-headers-$(uname -r)</nowiki>
	===NOTES===		===NOTES===
	This seems to need to be done sometimes after running updates on the host and rebooting.		This seems to need to be done sometimes after running updates on the host and rebooting.
Line 207:		Line 208:
	lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file		lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file
	</nowiki>		</nowiki>

	==Net Passthrough==		==Net Passthrough==
	This is important if your container needs to be able to create a TUN device. This is useful for applications like openvpn.		This is important if your container needs to be able to create a TUN device. This is useful for applications like openvpn.

Deathbybandaid: /* On Host */

2023-01-13T17:29:19Z

On Host

← Older revision		Revision as of 17:29, 13 January 2023
Line 105:		Line 105:

	<nowiki>reboot</nowiki>		<nowiki>reboot</nowiki>

			<nowiki>apt install -y build-essential</nowiki>

	[https://www.nvidia.com/Download/index.aspx Download Nvidia Driver for your card]		[https://www.nvidia.com/Download/index.aspx Download Nvidia Driver for your card]

Deathbybandaid: /* On Host */

2023-01-13T17:17:24Z

On Host

← Older revision		Revision as of 17:17, 13 January 2023
Line 94:		Line 94:
	===Nvidia===		===Nvidia===
	====On Host====		====On Host====


			Blacklist Nouveau Drivers

			<nowiki>nano /etc/modprobe.d/blacklist.conf</nowiki>

			<nowiki>blacklist nouveau</nowiki>

			<nowiki>update-initramfs -u</nowiki>

			<nowiki>reboot</nowiki>

	[https://www.nvidia.com/Download/index.aspx Download Nvidia Driver for your card]		[https://www.nvidia.com/Download/index.aspx Download Nvidia Driver for your card]
	<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>		<nowiki>chmod +x NVIDIA-Linux-x86_64-460.84.run</nowiki>
Line 151:		Line 163:
	lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file		lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file
	</nowiki>		</nowiki>

	====In Container====		====In Container====
	Copy the drivers over from the host somehow.		Copy the drivers over from the host somehow.

Deathbybandaid: /* Application Tricks for running in LXC */

2022-11-30T19:58:30Z

Application Tricks for running in LXC

← Older revision		Revision as of 19:58, 30 November 2022
Line 282:		Line 282:

	<nowiki>systemctl start apache2.service</nowiki>		<nowiki>systemctl start apache2.service</nowiki>

			==LXC elasticsearch==
			<nowiki>nano /etc/elasticsearch/jvm.options.d/heap.options</nowiki>

			<nowiki>
			-Xms4g
			-Xmx4g
			</nowiki>

			<nowiki>systemctl restart elasticsearch.service</nowiki>

	==Forward ALSA audio to LXC container==		==Forward ALSA audio to LXC container==
	===Host Machine===		===Host Machine===
Line 328:		Line 339:
	lxc.apparmor.profile: unconfined		lxc.apparmor.profile: unconfined
	</nowiki>		</nowiki>

	=Reduce the size of a container=		=Reduce the size of a container=
	Containers can always be easily increased in size via the web panel. Reducing their size is a slightly more complicated procedure that must be done via CLI.		Containers can always be easily increased in size via the web panel. Reducing their size is a slightly more complicated procedure that must be done via CLI.

Deathbybandaid: /* Docker */

2022-02-22T13:01:28Z

Docker

@@ Line 224: / Line 224: @@
 =Application Tricks for running in LXC=
 ==Docker==
-With the container stopped, Go to Options and edit the features. Enable CIFS and keyctl.
 Install
@@ Line 231: / Line 251: @@
   sh get-docker.sh
   </nowiki>
 ==Run NTP server in LXC==
   <nowiki>lxc.cap.drop = sys_module mac_admin mac_override</nowiki>