RDP Session Video Compression: Configuring Video-Worker

NOTE: This feature is available only in the Enterprise Edition

1. Introduction

Starting from JumpServer v3.0.0, the Video-Worker component was introduced. Its main purpose is to process recordings created by Razor (RDP Client) and Lion (Web RDP) and convert them into MP4 format for more convenient auditing and reduced disk space usage.

1.1. How It Works

• After startup, Video-Worker registers itself as a component (similar to Koko) to obtain the recording storage settings where the converted MP4 files will be uploaded.

• If Video-Worker is enabled in the Lion and Razor configuration, they upload their recordings directly to this component for transcoding.

• After successful transcoding, the MP4 file is uploaded to the configured storage.

If Lion or Razor cannot upload a recording to Video-Worker, they fall back to uploading it to their own storage.
When default storage is used, recordings are saved in

core/data/media/replay

Important Notes

• Video transcoding relies on ffmpeg, which heavily loads the CPU.

• CPU usage can reach 100%, which is why Video-Worker should NOT be deployed on the same machine as JumpServer.

• Razor: the recording is first converted into a series of images, and then ffmpeg compiles them into video — this consumes all available CPU cores.

• Lion (Web RDP): uses the official CLI tool guacenc, which fully loads a single CPU core during the entire process.

---

1.2. File Size Before and After Transcoding

---

1.3. Parameters

JumpServer URL used for API registration	Video-worker registers in core just like koko
BOOTSTRAP_TOKEN	none	Pre-generated secret key; must match JumpServer config
BIND_HOST	0.0.0.0	Bind IP
HTTPD_PORT	9000	HTTP/WS port
LOG_LEVEL	INFO	DEBUG / INFO / WARN / ERROR
VIDEO_WORKER_COUNT	0	Number of parallel workers; 0 = CPU cores + 1	If >0, the specified value is used
VIDEO_WORKER_SECRET_KEY	''	Random 32-character key for encryption/decryption	Must be the same for multiple workers
VIDEO_WORKER_CLEAN_PERIOD	7	Log retention period in days	Cleanup every 24 hours
IGNORE_VERIFY_CERTS	False	Ignore HTTPS certificate validation for CORE_HOST	Supported since v3.10.8+

---

2. Enabling Video-Worker on the Main JumpServer (Not Recommended)

JumpServer configuration file:

/opt/jumpserver/config/config.txt

Add the following parameters to enable Windows recording transcoding to MP4:

USE_VIDEO=1 
ENABLE_VIDEO_WORKER=True 
VIDEO_WORKER_HOST=http://jms_video:9000

Restart JumpServer:

jmsctl restart

Note:
Since ffmpeg heavily loads the CPU, running Video-Worker on the same node as JumpServer is not recommended. Consult JumpServer support for planning recommendations.

---

3. Deploying a Separate Transcoding Server

JumpServer version on the transcoding server must match the core environment version.

---

3.1. Deployment Using Docker Compose

Example Environment

---

Server Preparation

Unpack the JumpServer offline package and prepare the environment:

tar -zxvf jumpserver-offline-release-v3.10.9-amd64.tar.gz
cd /opt/jumpserver-offline-release-v3.10.9-amd64/scripts
./2_install_docker.sh      # установка Docker
docker load -i images/video-worker:v3.10.9.tar

---

Video-Worker Configuration

Edit:

vim /data/config.txt

Example config:

CORE_HOST=http://192.168.10.10:8080
BOOTSTRAP_TOKEN=YTc1NjNmNDItZjNlYS00NTMw   # должен совпадать с конфигом JumpServer

IGNORE_VERIFY_CERTS=true  # игнорировать сертификат HTTPS при подключении к CORE_HOST

VIDEO_WORKER_COUNT=6       # количество параллельных задач
VIDEO_WORKER_CLEAN_PERIOD=7  # период очистки логов (в днях)

HTTPD_PORT=9000
LOG_LEVEL=ERROR

---

Docker Compose file

vim /data/docker-compose-video.yml

(Full content preserved from original)

version: '2.4'

services:
  video:
    image: registry.fit2cloud.com/jumpserver/video-worker:v3.10.9
    container_name: jms_video
    hostname: jms_video
    cpus: 2
    ulimits:
      core: 0
    restart: always
    tty: true
    ports:
      - 9000:9000
    env_file:
      - /data/config.txt
    volumes:
      - /data/video/data:/opt/video/data
    healthcheck:
      test: "curl -fsL http://localhost:9000/video-worker/health/ > /dev/null"
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 10s
    networks:
      - video_net

---

JumpServer Configuration

vim /opt/jumpserver/config/config.txt

ENABLE_VIDEO_WORKER=true
VIDEO_WORKER_HOST=http://192.168.10.11:9000

---

Managing Video-Worker

docker-compose -f docker-compose-video.yml up -d   # запуск
docker-compose -f docker-compose-video.yml down    # остановка

---

3.2. Deployment Using Original JumpServer Files

Configure the Transcoding Server

Edit:

vim /opt/jumpserver/config/config.txt

Disable unnecessary components:

CORE_ENABLED=0
CELERY_ENABLED=0
KOKO_ENABLED=0
LION_ENABLED=0
MAGNUS_ENABLED=0
CHEN_ENABLED=0
KAEL_ENABLED=0
PANDA_ENABLED=0
WEB_ENABLED=0
XRDP_ENABLED=0

USE_VIDEO=1
ENABLE_VIDEO_WORKER=True
CORE_HOST=http://192.168.10.10:8080
BOOTSTRAP_TOKEN=YTc1NjNmNDItZjNlYS00NTMw

Then update the corresponding YAML files and restart:

cd /opt/jumpserver-offline-release-v3.10.9-amd64/compose/
jmsctl restart

---

JumpServer Side Configuration

vim /opt/jumpserver/config/config.txt

ENABLE_VIDEO_WORKER=true
VIDEO_WORKER_HOST=http://192.168.10.11:9000

Update YAML files:

cd /opt/jumpserver-offline-release-v3.10.9-amd64/compose/

And restart:

jmsctl restart

---

Diagnostics

If transcoding fails, check two possible areas:

1. Transcoding logs

Find the session log by session ID:

video/data/tasks/{id}

2. Lion and Razor logs

If there is no corresponding task entry, the issue may be related to file upload to Video-Worker.