Background

See also remote Docker server via SSH.

Basics

Reference: FFmpeg syntax and examples

Quick reference

Combining video and audio streams from different sources (assuming stream #0 for video and stream #1 for audio):

1

ffmpeg -i [input_source_0] -i [input_source_1] -c copy -map 0:v:0 -map 1:a:0 [output]

Streaming

Streaming capability

Be aware that not all the video container formats allow for streaming. Compared to the commonly used .mp4, effective streamable video container (muxer) names for use with FFmpeg are Matroska (.mkv), MPEG-TS (.m2ts preferred over .ts to be distinguished against TypeScript) and so on.

If the proper format option is missing, an error may occur from FFmpeg as:

1

Unable to choose an output format for 'pipe:'; use a standard extension for the filename or specify the format manually.

Preview an outbound stream in real-time replay

1

ffmpeg -i [input_source] -f [format] - | ffplay -f [format] -

Pipe and stream videos over SSH outbound to a remote ffmpeg process

Note: -y option is to skip any keyboard input prompt (in case of overriding an existing output file) to avoid the “broken pipe” error.

1

ffmpeg -i [input_source] -f [format] - | ssh [remote_user]@[remote_host] "ffmpeg -y -i pipe:0 [options] [output]"

Preview an inbound stream in real-time replay

1

ssh [remote_user]@[remote_host] "ffmpeg -i [input_source] -f [format] -" | ffplay -f [format] -

Pipe and stream videos over SSH inbound from a remote ffmpeg process

1

ssh [remote_user]@[remote_host] "ffmpeg -i [input_source] -f [format] pipe:1" | cat > output_file

Performance/Quality Tuning

Nvidia GPU-accelerated video processing

Reference 1
Reference 2
Using FFmpeg with NVIDIA GPU Hardware Acceleration

CUDA

1

ffmpeg -i [input] -hwaccel cuda [output]

CUVID

1

ffmpeg -i [input] -c:v h264_cuvid [output]

Full hardware transcode with NVDEC and NVENC

1

ffmpeg -hwaccel cuda -hwaccel_output_format cuda -i [input] -c:v h264_nvenc [output]

Further extension - batch operation

Batch video processing

Background

There appears to be lots of screen recordings that occupy gigabytes of storage space locally on an MBP, which need to be compressed. While the local tasks running on an MBP may cause the heat dissipating fan to roaring, making use of an idle desktop seems a viable solution to accelerate the workflow.

Considerations Before Starting

• Q: Why not refer to powerful cloud-top computing resources such as those on AWS/GCP/Azure?
• A: Extra costs will apply, yet the deploying, debugging and tuning operations can’t be skipped still. On the contrary, a personal desktop as the server host is likely more cost-effective and easily maintainable for personal use cases versus in production.
• Q: Why not use a U-disk to transfer the files from the MBP to the desktop device?
• A: The source files are gigabytes of videos that can’t fit into the U-disk given the capacity limit.
• Q: Why not establish a simple web service on the host over HTTP for a user-friendly interface?
• A: Those web UI features (see also Docker API documentation) are more than necessary to achieve the goal. In the meantime, exposing the port 2375 (reference) over HTTP is not secured, yet the HTTPS/TLS setup via port 2376 introduced lots of complexity even when following the Docker’s documentation.

Quick Start

Counter-intuitively, there’s actually no need to install or run Docker Desktop app on either the client or server side, as long as the Docker daemon is ready on the server host and Docker CLI is present on the client side, saving lots of confusion, time and effort on the unnecessary desktop app settings.

Either Docker daemon and Docker CLI can be installed on the common *nix OS or distros including MacOS. If on Windows, Windows Subsystem Linux (WSL) will be helpful to prepare the runtime environment without the need to reinstall the entire OS or run a virtual machine, by enabling the certain optional feature instead.

Even if Docker Desktop is installed on Windows, the runtime environment has to be held in WSL anyway.

Server-side setup in Linux/WSL distro of Ubuntu (as an example)

• Package sources

  Common practice to refresh the list of available packages and their versions:

  1	sudo apt update

  Additionally, set up Docker’s apt repository using Docker’s official GPG key and Apt sources (recommended; see also Install Docker Engine on Ubuntu).

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

  # Add Docker's official GPG key: sudo apt update sudo apt install ca-certificates curl sudo install -m 0755 -d /etc/apt/keyrings sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc sudo chmod a+r /etc/apt/keyrings/docker.asc # Add the repository to Apt sources: sudo tee /etc/apt/sources.list.d/docker.sources <<EOF Types: deb URIs: https://download.docker.com/linux/ubuntu Suites: $(. /etc/os-release && echo "${UBUNTU_CODENAME:-$VERSION_CODENAME}") Components: stable Signed-By: /etc/apt/keyrings/docker.asc EOF sudo apt update

• Where is ifconfig?

  After updating, install the net-tools package, which contains ifconfig, using this command:

  1	sudo apt install net-tools

• OpenSSH server

  1	sudo apt install openssh-server ssh-askpass ufw

  Enable, Start and check SSH Service:

  1 2 3

  sudo systemctl enable ssh sudo systemctl start ssh sudo systemctl status ssh

  Adjust Firewall (using UFW):

  1	sudo ufw allow ssh

• Docker engine, daemon and container management (latest version)

  1	sudo apt install docker-ce docker-ce-cli containerd.io

  The Docker service starts automatically after installation. To verify that Docker is running, use:

  1	sudo systemctl status docker

  Verify that the installation is successful by running the hello-world image (expecting some human-readable self-explanatory outputs):

  1	sudo docker run hello-world

• More usages about Docker daemon (read more)

  Test Docker daemon connection in a debug mode:

  1 2 3 4 5

  dockerd --debug [--tls=true \] [--tlscert=/var/docker/server.pem \] [--tlskey=/var/docker/serverkey.pem \] [--host tcp://192.168.59.3:2376]

  By default the daemon stores data in:

  • /var/lib/docker on Linux
  • C:\ProgramData\docker on Windows

Cliend-side setup (reference)

Docker CLI needs to be installed on the client side as well, same as above.

Create a Docker context to specify a non-local Docker host:

1
2
3
4

docker context create \
    --docker host=ssh://docker-user@host1.example.com \
    --description="Remote engine" \
    my-remote-engine

The context doesn’t take effect immediately upon creation. The next step is to inform the Docker CLI to use it, and to connect to the remote engine:

1

docker context use my-remote-engine

Quickly test if the context works out:

1

docker info

Details

Additional SSH usages

• Executing a simple command:

1

ssh user@example.com "ls -l /home/user"

Windows environment setup (skip this if not using Windows on a host)

Before digging deeper, beware of the difference between the concepts of WSL and WSL distros (read more).

What’s more, the current versions of WSL are 2.x, also referred to as WSL2. And the WSL command line tool will run in 2.x version by default.

WSL Distros

• List all the distro versions ready to install

  1	wsl --list --online

• List the local distros installed

  1	wsl --list [--verbose]

• Install a specific distro (e.g. Ubuntu)

  1	wsl --install ubuntu

• Start a WSL distro instance - simply run wsl in the built-in Command Prompt, Terminal App or Powershell.

• Check for the private IP address assigned to the Distro, which is helpful for the upcoming network configs

  1	wsl hostname -I

• Stop a running WSL distro only without removing its filesystem

  As an example, it will make the changes to the .wslconfig file take effect:

  1	wsl --shutdown

• Remove a WSL distro instance with its filesystem

  1	wsl --unregister ubuntu

• Export a WSL distro to a file

  It helps to create a backup (as .tar file by default) or relocate the distro to a different directory (shutdown recommended as above):

  1	wsl --export <DistributionName> <Full path to .tar/.vhdx FileName> [--vhd]

• Import a file as a WSL distro

  This allows to create a distro filesystem in a different directory other than before:

  1	wsl --import <NewDistributionName> <InstallLocation> <Full path to .tar/.vhdx FileName>

Network Configs on a Windows host

• Allow for incoming ping requests

  Add this rule to the firewall via Powershell with administrator permissions to resolve the timeout error of ping:

  1	New-NetFirewallRule -Name "Ping" -Protocol ICMPv4 -Direction Inbound -Action Allow

• Allow for incoming SSH requests

  This is to enable the built-in firewall rule for the OpenSSH Server:

  1	Enable-NetFirewallRule -DisplayGroup "OpenSSH Server"

  Alternatively, manually create a new inbound firewall rule:

  1	New-NetFirewallRule -DisplayName "Allow SSH Inbound" -Direction Inbound -Protocol TCP -LocalPort 22 -Action Allow -Profile Any

• Set up Port Forwarding on Windows host

  This is to add a port proxy rule to forward a local port 22 on Windows host to the WSL SSH server port 22 in WSL:

  1	netsh interface portproxy add v4tov4 listenport=22 listenaddress=0.0.0.0 connectport=22 connectaddress=<wsl-ip>

• Test connecting to WSL server via SSH from another device

  1	ssh username@host_ip [-p 22]

• Alternative: SSH Tunnels from Windows to a Remote Host via WSL (SSH Tunneling)

  This option hasn’t been tested and verified yet:

  1	ssh -L <local-port>:localhost:<remote-port> <remote-host>

Miscallaneous

Known Issues of WSL

• Why does WSL end, and the established SSH connection get closed, after a while of being idle?

  The cause lies in the WSL Idle Timeout setting. By default, WSL 2 goes into a suspended state (after a short period) once all open terminal windows and running processes have exited. The best way to prevent this is to configure the vmIdleTimeout setting in a Windows configuration file.

  Navigate to the Windows user profile directory: %userprofile% (e.g., C:\Users\YourUsername).
  Create a file named .wslconfig if it doesn’t exist.
  Open the file and add the following lines:

  1 2 3 4

  [wsl2] ; Sets the time in seconds that a WSL 2 distro can be idle before it is terminated. ; To disable the timeout, set this to -1. vmIdleTimeout=-1

Known Issue of OpenSSH / Docker

• docker stderr=ssh_askpass: exec(/usr/bin/ssh-askpass): No such file or director

  External reference

  Install the ssh-askpass package:

  1	sudo apt install [-y] ssh-askpass

• “Connection timeout” even with SSH connection tested okay

Check docker context on both host and client sides:

1
2
3

docker context ls
docker context show
docker context use [context name]

Ever since 2025, there has been an increasing number of in-depth problems involved in the seemingly resolvable coding interviews, which used to be very uncommon. It’s worthwhile noting that being unprepared for such topics but merely showing a problem-solving mindset may not always lead to a desirable passing-level result, given the extremely-competitive job market, while it’s also not recommended to learn by rote - memorizing the answer without thorough understanding could be easily recognized as imposter by the interviewers.

Discord Screen Challenge - Chat Server

Problem Statement

Original Problem Statement

Functional Requirements:

• The chat server should be capable of handling multiple clients connecting at the same time.
• Disconnection should be handled in a clean way.
• Should be able to receive and forward messages among clients (message is not echoed back to sender).

The questions below will be in a reverse-chronological order as they appeared in the interviews.

Simulated Calculator in String Processing

Calculator V1 - LC 224. Basic Calculator (Hard)

Essentially, merely +- arithmetic operations and parenthesis (nested possible) need to be supported.

Concise & efficient Solution after refactoring

Upon second thought, it appears that stack operations are need only when destructing parenthesis and processing the +- operators (see also other shared solutions on LeetCode).

Uber 60-min FE Interview - mid-July

Problem: on Hackerrank online IDE, given the function definitions, implement certain logic to obtain the expected output while ensuring the function calls running in parallel.

Note that (based on clarification made by the interviewer):

• Only JS/TS allowed for this problem;
• The function of getNameById is meant to simulate certain time consuming task; no change should be made other than under TODO inside asyncMap;
• callback param passed into getNameById is not the same as that in asyncMap.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

function getNameById(id, callback) {
  // simulating async request
  const randomRequestTime = Math.floor(Math.random() * 100) * 200;
  setTimeout(() => {
    callback("User" + id);
  }, randomRequestTime);
}

// Input
const userIds = [1, 2, 3, 4, 5];
async function asyncMap(input, iterateeFn, callback) {
  // TODO: implement this function, ensuring parallelism and the correct ordering in the output
}

asyncMap(userIds, getNameById, (names) => {
  console.log(names); // * ["user1", "user2", "user3", "user4", "user5"]
});

These usages are helpful to build a solution of NAS Status Monitor Visualization.

Fix pacman-mirrors: ModuleNotFoundError

Look for the missing modules failed to be imported:

1

sudo pacman -S python3 python-urllib3 python-charset-normalizer python-idna

Fix blocking pacman upgrade failure

Option 1:

1
2

sudo pacman-mirrors -g -f5
pamac upgrade --force-refresh

Option 2:

1

sudo pacman-mirrors --fasttrack && sudo pacman -Syyu

Also check for

1

yay -Suy

It’s sometimes possibly also needs to remove conflicting packages to proceed, rather than trying to install by force using the pacman options -d/-dd/-nodeps.

Allow installing dependencies from AUR

1

sudo pacman -S yay

UI tools

1
2
3

yay -S gotop
sudo pacman -S bspwm sxhkd polybar picom dunst dmenu nitrogen
sudo pacman -S mpd ncmpcpp

UI modules

1
2

sudo pacman -S rofi calc xorg xorg-xinit xorg-xbacklight alsa-utils
yay -S pywal networkmanager-dmenu

Font dependencies

1
2

sudo pacman -S ttf-font-awesome [ttf-dejavu]
yay -S ttf-material-design-iconic-font ttf-unifont ttf-cascadia-code ttf-nerd-fonts-symbols ttf-material-icons

Check if TTF fonts are present in the OS:

1
2

fc-cache -f -v
fc-list | grep ...

Clock

1

systemctl start mpd.service

Press = after starting ncmpcpp in CLI.

LRU Cache

1
2
3
4
5
6

class Node:
    def __init__(self, key, value):
        self.prev = None
        self.next = None
        self.key = key
        self.val = value

Dependencies

• Issue of dependency conflict or failure to find the dependency that should be there:

  sudo pacman -Syu –debug
  It triggers a prompt
  This also performs wide-range package upgrades.

• Issue of CMake / makepkg -si:

  1 2 3 4 5 6 7 8

  ==> Starting build()... CMake Error: CMake was unable to find a build program corresponding to "Unix Makefiles". CMAKE_MAKE_PROGRAM is not set. You probably need to select a different build tool. CMake Error: CMAKE_C_COMPILER not set, after EnableLanguage CMake Error: CMAKE_CXX_COMPILER not set, after EnableLanguage -- Configuring incomplete, errors occurred! ==> ERROR: A failure occurred in build(). Aborting...

  To resolve such building depedencies:

  sudo pacman -S base-devel

  1 2 3 4 5

  Packages (12) autoconf-2.71-4 automake-1.16.5-2 bison-3.8.2-5 debugedit-5.0-5 flex-2.6.4-5 gc-8.2.2-1 guile-3.0.9-1 m4-1.4.19-3 make-4.4.1-2 patch-2.7.6-9 pkgconf-1.8.1-1 base-devel-1-2

• Install pytorch with ROCm backend (read more)

  For CPUs with AVX2 instruction set support, that is, CPU microarchitectures beyond Haswell (Intel, 2013) or Excavator (AMD, 2015), install python-pytorch-opt-rocm to benefit from performance optimizations. Otherwise install python-pytorch-rocm.

  sudo pacman -S python-pytorch-opt-rocm

  This may generate an installation as large as:

  1 2 3 4

  Total Download Size: 2654.00 MiB Total Installed Size: 29579.85 MiB :: Proceed with installation? [Y/n]

Appearance

• Icon font not displaying in Code terminal

  sudo pacman -S ttf-meslo-nerd-font-powerlevel10k

  1 2 3 4 5

  { ... "terminal.integrated.fontFamily": "MesloLGS NF", ... }

Preparation

A ZFS pool should surely pre-exist there to allow for the operations introduced later.

• Key point: Why not split, or shrink existing volumes for a Time Machine volume?

  Not viable. A ZFS partition always tries to occupy as much space as possible, so does ZFS volumes under it, sharing the entire space of it. To reserving some space is exactly what the quota property is designed for.

  In addition, you can use the reservation property to guarantee that a specified amount of disk space is available to a file system. Both properties apply to the dataset on which they are set and all descendents of that dataset.

  More…

• Creating a new ZFS filesystem

  zfs create -o compression=lz4 pool/filesystem

  where the purpose of an optional argument -o compression=lz4 is to turn on the compression for this dataset.

Unordered MultiSet

1
2
3
4
5
6
7
8
9
10
11
12
13

class MultiSet:
    def __init__(self):
        self.data = dict()
    
    def add(self, val):
        self.data[val] = 1 + (self.data[val] if val in self.data else 0)
    
    def remove(self, val):
        if not val in self.data:
            return
        self.data[val] -= 1
        if self.data[val] == 0:
            del self.data[val]

Fenwick Tree / Binary-indexed Tree

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

class BIT:
    def __init__(self, size):
        self.size = size
        self.data = [0] * size
    
    def add(self, pos, val):
        while pos < self.size:
            self.data[pos] += val
            pos += pos & (-pos)
    
    def query(self, pos):
        res = 0
        while pos > 0:
            res += self.data[pos]
            pos &= pos - 1
        return res