serhii.net

In the middle of the desert you can say anything you want

30 Apr 2021

Day 850

grep ignore case

lspci | grep -i "nvidia"

-i == ‘ignore case’ is actually something that I can remember.

Docker (stop) autostart of container

Docker will autostart any container with a RestartPolicy of ‘always’ when the docker service initially starts. 1

I can set/unset it in kitematic, or through terminal:

docker update --restart=no my-container

apt-get purge remove –autoremove etc

Quoting SO: 2

    apt purge --auto-remove <packagename>

purges packagename and any packages which are rendered unnecessary by its removal, as well as any other packages which aren’t necessary.

    apt autoremove --purge

purges any packages which aren’t necessary (marked as “automatically installed” and with no dependent packages).

The first form is what you’d use when manipulating individual packages; the latter is a clean-up operation across all packages.

Ways to clean up with apt-get - tutorial

This seems nice, TODO: Cleaning up with apt-get | Network World

Backing up LVM disk encryption keys

LVM - Debian Wiki is nice and readable. I used this command to backup the headers:

 sudo cryptsetup luksHeaderBackup /dev/nvmeXXXXX   --header-backup-file headerBackupFile

… and put it somewhere not on the drive I’ll be recovering if it all goes wrong.

Setting up Tensorflow and CUDA with an eGPU

Aaaand the saga continues!

…since the GPU is an eGPU, apparently I do need to do the harder way: Accelerating Machine Learning on a Linux Laptop with an External GPU | NVIDIA Developer Blog

Getting the eGPU detected

It is, I can see it:

(17:42:42/10815)~/$ lspci | grep -i VGA
00:02.0 VGA compatible controller: Intel Corporation UHD Graphics 620 (rev 07)
0c:00.0 VGA compatible controller: NVIDIA Corporation GP104 [GeForce GTX 1070] (rev a1)

but if it wasn’t, I’d authorize it and check with boltctl list:

(17:43:13/10817)~/$ boltctl list
[...]
 ● GIGABYTE GV-N1070IXEB-8GD
   ├─ type:          peripheral
   ├─ name:          GV-N1070IXEB-8GD
   ├─ vendor:        GIGABYTE
   ├─ uuid:          # redacted
   ├─ status:        authorized
   │  ├─ domain:     domain0
   │  └─ authflags:  none
   ├─ authorized:    Do 29 Apr 2021 07:57:37 UTC
   ├─ connected:     Do 29 Apr 2021 07:57:37 UTC
   └─ stored:        no

How to setup an eGPU on Ubuntu for TensorFlow describes other things that can go wrong:

I had to disable the following, otherwise my eGPU was not detected:

  • Secure Boot
  • Thunderbolt Security Level

From this point on, I follow Nvidia’s tutorial 3 unless stated otherwise.

Purging, cleaning up old broken install attempts, updating and upgrading

Using quotes means the * doesn’t have to be escaped.

sudo apt-get purge "nvidia*"

This is a fuller example: 4

sudo rm /etc/apt/sources.list.d/cuda*
sudo apt remove --autoremove nvidia-cuda-toolkit
sudo apt remove --autoremove nvidia-*

Found and manually removed /etc/apt/sources.list.d/graphics-drivers-ubuntu-ppa-bionic.list, leaving the .save file in place.

As per nvidia’s guide,

sudo apt-get update
sudo apt-get dist-upgrade

To be safe, rebooted.

Downloading the correct drivers

The existing driver is most likely Nouveau, an open-source driver for NVIDIA GPUs. Because Nouveau doesn’t support eGPU setups, install the NVIDIA CUDA and NVIDIA drivers instead. You must also stop the kernel from loading Nouveau. 3

okay!

Change of plan - what is NVIDIA data-science-stack?

Found this: NVIDIA/data-science-stack: NVIDIA Data Science stack tools Read about it here: Ubuntu for machine learning with NVIDIA RAPIDS in 10 min | Ubuntu

Official by nvidia, and seems to do automatically what’s needed for supported systems. Let’s run a script from the internet that installs drivers, loads kernel modules etc.

Source is available, yay for open source: data-science-stack/data-science-stack at master · NVIDIA/data-science-stack

Ran ./data-science-stack setup-system - uses sudo, didn’t ask for root or anything.o

Seems to have installed nvidia driver version 460. Asked to reboot at the end.

Rebooted.

(18:40:30/10909)~/$ nvidia-smi
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

okay. Same results I had. Confirms that my prev. steps weren’t wronger than the script.

(18:41:49/10910)~/$ sudo apt list --installed | grep "\(cuda\|nvidia\)"

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

libnccl2/unknown,now 2.9.6-1+cuda11.3 amd64 [installed]
libnvidia-cfg1-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-common-460/unknown,now 460.73.01-0ubuntu1 all [installed,automatic]
libnvidia-compute-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-container-tools/bionic,now 1.4.0-1 amd64 [installed,automatic]
libnvidia-container1/bionic,now 1.4.0-1 amd64 [installed,automatic]
libnvidia-decode-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-encode-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-extra-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-fbc1-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-gl-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
libnvidia-ifr1-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
nvidia-compute-utils-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
nvidia-container-runtime/bionic,now 3.5.0-1 amd64 [installed,automatic]
nvidia-container-toolkit/bionic,now 1.5.0-1 amd64 [installed,automatic]
nvidia-dkms-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
nvidia-docker2/bionic,now 2.6.0-1 all [installed]
nvidia-driver-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed]
nvidia-kernel-common-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
nvidia-kernel-source-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
nvidia-prime/bionic-updates,bionic-updates,now 0.8.16~0.18.04.1 all [installed,automatic]
nvidia-settings/unknown,unknown,now 465.19.01-0ubuntu1 amd64 [installed,automatic]
nvidia-utils-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]
xserver-xorg-video-nvidia-460/unknown,now 460.73.01-0ubuntu1 amd64 [installed,automatic]

Also, as usual,

(18:48:34/10919)~/$ lsmod | grep nvi
(18:48:37/10920)~/$

lspci -k shows the kernel modules:

0c:00.0 VGA compatible controller: NVIDIA Corporation GP104 [GeForce GTX 1070] (rev a1)
        Subsystem: Gigabyte Technology Co., Ltd GP104 [GeForce GTX 1070]
        Kernel modules: nvidiafb, nouveau

This output implies no nvidia driver is installed on my system5. …though it is.

$ nvidia-settings --version
nvidia-settings:  version 465.19.01

software-properties-gtk tells me I’m using the proprietary nvidia-driver-460, not 465

In any case, can’t blacklist nouveau as still there are no ubuntu kernel modules.

BUT!

(19:04:04/10946)~/$ dkms status
nvidia, 460.73.01: added

Also, inxi -Fxxxrz (found somewhere on the internet):

Graphics:  Card-1: Intel UHD Graphics 620 bus-ID: 00:02.0 chip-ID: 8086:5917
           Card-2: NVIDIA GP104 [GeForce GTX 1070] bus-ID: 0c:00.0 chip-ID: 10de:1b81
           Display Server: x11 (X.Org 1.19.6 ) drivers: modesetting,nvidia (unloaded: fbdev,vesa,nouveau)

It it sees them as there and loaded? Does dkms somehow bypass lsmod etc?

sudo dkms autoinstall should autoinstall all added drivers, …let’s hope for the best I guess.

(19:11:47/10958)~/$ sudo dkms autoinstall

Kernel preparation unnecessary for this kernel.  Skipping...
applying patch disable_fstack-clash-protection_fcf-protection.patch...patching file Kbuild
Hunk #1 succeeded at 85 (offset 14 lines).


Building module:
cleaning build area...
unset ARCH; [ ! -h /usr/bin/cc ] && export CC=/usr/bin/gcc; env NV_VERBOSE=1 'make' -j8 NV_EXCLUDE_BUILD_MODULES='' KERNEL_UNAME=5.4.0-72-generic IGNORE_XEN_PRESENCE=1 IGNORE_CC_MISMATCH=1 SYSSRC=/lib/modules/5.4.0-72-generic/build LD=/usr/bin/ld.bfd modules......(bad exit status: 2)
ERROR: Cannot create report: [Errno 17] File exists: '/var/crash/nvidia-dkms-460.0.crash'
Error! Bad return status for module build on kernel: 5.4.0-72-generic (x86_64)
Consult /var/lib/dkms/nvidia/460.73.01/build/make.log for more information.

The file is long, keys seems:

 scripts/Makefile.build:269: recipe for target '/var/lib/dkms/nvidia/460.73.01/build/nvidia/nv.o' failed
 make[2]: *** [/var/lib/dkms/nvidia/460.73.01/build/nvidia/nv.o] Error 1
 Makefile:1754: recipe for target '/var/lib/dkms/nvidia/460.73.01/build' failed
 make[1]: *** [/var/lib/dkms/nvidia/460.73.01/build] Error 2
 make[1]: Leaving directory '/usr/src/linux-headers-5.4.0-72-generic'
 Makefile:80: recipe for target 'modules' failed
 make: *** [modules] Error 2
DKMSKernelVersion: 5.4.0-72-generic
Date: Fri Apr 30 18:30:45 2021
DuplicateSignature: dkms:nvidia-dkms-460:460.73.01-0ubuntu1:/var/lib/dkms/nvidia/460.73.01/build/conftest/functions.h:11:2: error: #error acpi_walk_namespace() conftest failed!
Package: nvidia-dkms-460 460.73.01-0ubuntu1
PackageVersion: 460.73.01-0ubuntu1
SourcePackage: nvidia-graphics-drivers-460
Title: nvidia-dkms-460 460.73.01-0ubuntu1: nvidia kernel module failed to build

Smells like a driver/kernel support isse?

First result when googling dkms nvidia 460 is this: Can’t get nvidia 460 module to build on Ubuntu 20.04 to support two A100s - GPU Unix Graphics / Linux - NVIDIA Developer Forums

Please check if the build symlink to the headers for dkms exists:

ls /lib/modules/$(uname -r)/build

Otherwise, create it

ln -s /usr/src/linux-headers-$(uname -r)  /lib/modules/$(uname -r)/build

Didn’t have it, created it, trying again, same error, deleted the previous log, full output is:

(19:19:54/10967)~/$ sudo dkms autoinstall

Kernel preparation unnecessary for this kernel.  Skipping...
applying patch disable_fstack-clash-protection_fcf-protection.patch...patching file Kbuild
Hunk #1 succeeded at 85 (offset 14 lines).


Building module:
cleaning build area...
unset ARCH; [ ! -h /usr/bin/cc ] && export CC=/usr/bin/gcc; env NV_VERBOSE=1 'make' -j8 NV_EXCLUDE_BUILD_MODULES='' KERNEL_UNAME=5.4.0-72-generic IGNORE_XEN_PRESENCE=1 IGNORE_CC_MISMATCH=1 SYSSRC=/lib/modules/5.4.0-72-generic/build LD=/usr/bin/ld.bfd modules.......(bad exit status: 2)
Error! Bad return status for module build on kernel: 5.4.0-72-generic (x86_64)
Consult /var/lib/dkms/nvidia/460.73.01/build/make.log for more information.

The file is full of what looks like syntax errors..?

This charming chinese website seems to imply gcc version is to blame: NVIDIA驱动出错:NVIDIA-SMI has failed because it couldn‘t communicate with the NVIDIA driver. Make sure t_sazass的博客-CSDN博客

(19:22:39/10974)~/$ cat /proc/version
Linux version 5.4.0-72-generic (buildd@lgw01-amd64-021) (gcc version 7.5.0 (Ubuntu 7.5.0-3ubuntu1~18.04)) #80~18.04.1-Ubuntu SMP Mon Apr 12 23:26:25 UTC 2021
sudo apt install gcc-8
sudo update-alternatives --config gcc
sudo update-alternatives --remove-all gcc
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-8 10
sudo update-alternatives --install /usr/bin/cc cc /usr/bin/gcc-8 10

Let’s retry dkms autoinstall:

(19:26:03/10981)~/$ sudo dkms autoinstall

Kernel preparation unnecessary for this kernel.  Skipping...
applying patch disable_fstack-clash-protection_fcf-protection.patch...patching file Kbuild
Hunk #1 succeeded at 85 (offset 14 lines).


Building module:
cleaning build area...
unset ARCH; [ ! -h /usr/bin/cc ] && export CC=/usr/bin/gcc; env NV_VERBOSE=1 'make' -j8 NV_EXCLUDE_BUILD_MODULES='' KERNEL_UNAME=5.4.0-72-generic IGNORE_XEN_PRESENCE=1 IGNORE_CC_MISMATCH=1 SYSSRC=/lib/modules/5.4.0-72-generic/build LD=/usr/bin/ld.bfd modules...............
Signing module:
 - /var/lib/dkms/nvidia/460.73.01/5.4.0-72-generic/x86_64/module/nvidia-modeset.ko
 - /var/lib/dkms/nvidia/460.73.01/5.4.0-72-generic/x86_64/module/nvidia.ko
 - /var/lib/dkms/nvidia/460.73.01/5.4.0-72-generic/x86_64/module/nvidia-uvm.ko
 - /var/lib/dkms/nvidia/460.73.01/5.4.0-72-generic/x86_64/module/nvidia-drm.ko
Secure Boot not enabled on this system.
cleaning build area...

DKMS: build completed.

nvidia.ko:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/5.4.0-72-generic/updates/dkms/

nvidia-modeset.ko:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/5.4.0-72-generic/updates/dkms/

nvidia-drm.ko:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/5.4.0-72-generic/updates/dkms/

nvidia-uvm.ko:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/5.4.0-72-generic/updates/dkms/

depmod...

DKMS: install completed.

WOW. WOOOOOW. WOOOOOOOOOOOOOOOOOOOOOO

Without even restarting, after the first command my screen flashed and changed resolution a bit, BUT THEN IT WORKED

(19:34:17/10983)~/$ nvidia-smi
No devices were found
(19:34:20/10984)~/$ nvidia-smi
Fri Apr 30 19:34:22 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.73.01    Driver Version: 460.73.01    CUDA Version: 11.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  GeForce GTX 1070    On   | 00000000:0C:00.0 Off |                  N/A |
|  0%   54C    P0    37W / 151W |      7MiB /  8119MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

All these attempts failed because the nvidia module in dkms couldn’t install because syntax errors because old gcc compiler version.

What could I have done differently? Why at no point did I see errors about the kernel module failing to build, where should I have looked for them? And why syntax errors instead of something checking the used gcc version and loudly failing when there was a mismatch? Why is that chinese website the only place I found this fix?

(19:42:57/10995)~/$ lsmod | grep nvidia
nvidia_uvm           1015808  0
nvidia_drm             57344  1
nvidia_modeset       1228800  1 nvidia_drm
nvidia              34123776  17 nvidia_uvm,nvidia_modeset
drm_kms_helper        188416  2 nvidia_drm,i915
drm                   491520  15 drm_kms_helper,nvidia_drm,i915

Now let’s hope this survives a restart. And that it works when the eGPU is disconnected.

NVIDIA data-science-stack

Following the readme, ran both options in separate terminals:

./data-science-stack list
./data-science-stack build-container
./data-science-stack run-container

and

./data-science-stack list
./data-science-stack build-conda-env
./data-science-stack run-jupyter

The latter seems to be installing CUDA and friends on my computer - didn’t expect it, but I need them either way I think, I guess I’ll let the script handle everything since it started. It installed conda to ~/conda/, but again, not sure what I was expecting

Both running for 20+ minutes now

EDIT: ~/conda/ took 20gb filling up my drive, blocking everything, deleted it

The docker with jupyterlab - tensorflow can’t access the GPU, but pytorch can.

Carrying on with setting the eGPU up

The NVIDIA eGPU tutorial3 continues with offloading Xorg to the GPU - do I want this? Can I use the GPU just for training, and leave Xorg running on the internal one? I probably don’t

Restarting and testing

As I remember from the last time, X doesn’t start when the GPU is connected at boot but everything’s fine when it gets connected after starting X. When it’s connected, it seems the driver gets loaded and nvidia-smi etc works. That the system works without the eGPU attached is nice! Plug-and-play is nice too.

Installed pytorch in a virtualenv, for cuda 11.1, test snippet says cuda works!

import torch
x = torch.rand(5, 3)
print(x)

torch.cuda.is_available()

Tensorflow:

>>> import tensorflow as tf
2021-04-30 21:36:12.984883: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0
>>> tf.debugging.set_log_device_placement(True)
>>> a = tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
2021-04-30 21:36:23.055614: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set
2021-04-30 21:36:23.058062: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcuda.so.1
2021-04-30 21:36:23.115366: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:941] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-04-30 21:36:23.116510: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 0 with properties:
pciBusID: 0000:0c:00.0 name: GeForce GTX 1070 computeCapability: 6.1
coreClock: 1.721GHz coreCount: 15 deviceMemorySize: 7.93GiB deviceMemoryBandwidth: 238.66GiB/s
2021-04-30 21:36:23.116553: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0
2021-04-30 21:36:23.119974: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcublas.so.11
2021-04-30 21:36:23.120034: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcublasLt.so.11
2021-04-30 21:36:23.121503: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcufft.so.10
2021-04-30 21:36:23.121842: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcurand.so.10
2021-04-30 21:36:23.125037: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcusolver.so.10
2021-04-30 21:36:23.125803: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcusparse.so.11
2021-04-30 21:36:23.125980: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory
2021-04-30 21:36:23.125996: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1757] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.

Which libcudnn?

Tensorflow’s tutorial (GPU support  |  TensorFlow) does this:

Install development and runtime libraries (~4GB)
sudo apt-get install --no-install-recommends \
    cuda-11-0 \
    libcudnn8=8.0.4.30-1+cuda11.0  \
    libcudnn8-dev=8.0.4.30-1+cuda11.0

What is the version for CUDA 11.2? cuDNN Archive | NVIDIA Developer has download links. The one for 11.2 is called “cudnn-11.2-linux-x64-v8.1.1.33.tgz”. I plug those versions in, they exist and install fine:

sudo apt-get install   libcudnn8=8.1.1.33-1+cuda11.2
sudo apt-get install   libcudnn8-dev=8.1.1.33-1+cuda11.2

And tensorflow now works!

2021-04-30 21:42:46.176942: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1406] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 7440 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1070, pci bus id: 0000:0c:00.0, compute capability: 6.1)

I can’t believe it but wow. It’s finished, it works, X didn’t die, plug-and-play works, no manual driver loading.

All in all, including all the failed attempts, took 5:30h of pure time, according to my time tracking.

The only wrinkle is that X doesn’t start when turning the computer on with the eGPU attached, but I can 100% live with that!

GPU benchmarking linux

How to Benchmark your GPU on Linux has a fun quote:

This tool is very old, very basic and only tests a small portion of today’s OpenGL capabilities. Back in the old days, it was used to determine if the proprietary driver was installed and running properly as open-source drivers were performing awfully enough to be perfectly noticeable during this test. Nowadays, you won’t notice any difference between the two

qutebrowser open a private window

Added this to config.py:

config.bind('<Alt-P>', 'set-cmd-text -s :open -p ')

Managing dotfiles with machine-specific configuration

Qutebrowser import other config files

Seen in someone’s config.py on gitlab6:

for f in glob.glob(str(config.configdir / 'conf.d/*.py')):
    config.source(str(os.path.relpath(f, start=config.configdir)))

Random i3 configs

Nice examples: i3_config/settings.d at master · kiddico/i3_config · GitHub

i3 doesn’t have any kind of include directive in the config files, sadly. i3 - Source/import file from i3wm config - Stack Overflow is one option:

bindsym $mod+Shift+c exec "cat ~/.config/i3/colors ~/.config/i3/base > ~/.config/i3/config && i3-msg reload"

A keybinding to overwrite the config file and restart i3 with a command.

To read - life hacking

This looks very interesting, I shouldn’t forget to go through this: Life Hacking His blog with personal examples: Alex Vermeer — Life-Hacking. Climbing. Striving for awesome. Coffee. — Page 2

A non-pdf description of Life Areas with questions and metrics for each.

(He’s the same guy who created the awesome How to Get Motivated: A Guide for Defeating Procrastination poster!)

And let’s remember the classic: Evidence-based advice on how to be successful in any job - 80,000 Hours

Detach process completely from terminal

Two options I like:7

  • nohup cmd &
  • cmd & disown

I feel one of these will become part of many aliases of mine.

And short bash function from the same place:

function dos() {
    # run_disowned and silenced

    run_disowned "$@" 1>/dev/null 2>/dev/null
}

  1. Docker – Prevent Container Autostart | eknori.de↩︎

  2. debian - What’s the right way to purge recursively with apt? - Unix & Linux Stack Exchange ↩︎

  3. Accelerating Machine Learning on a Linux Laptop with an External GPU | NVIDIA Developer Blog ↩︎

  4. CUDA 10.1 installation on Ubuntu 18.04 LTS | Medium ↩︎

  5. Install the Latest Nvidia Linux Driver - LinuxConfig.org ↩︎

  6. ~pvsr/dotfiles: qutebrowser/.config/qutebrowser/config.py - sourcehut git ↩︎

  7. linux - How do I detach a process from Terminal, entirely? - Super User ↩︎

Nel mezzo del deserto posso dire tutto quello che voglio.