Replay Worker Deployment
1. Replay Agent Installation
1.1 File Preparation
Copy the following to the target server:
- replay-agent.yml
- ms-replay.tar
Extract the archive and update Git credentials in replay-agent.yml.
Credentials are not encrypted in the current version.
1.2 Vector Dependency
Replay Agent launches an additional container using the latest Vector image:
docker pull timberio/vector:latest-alpine
For offline environments, save and load the image manually.
1.3 GitLab Agent Repository
In GitLab:
- Create a new project named agent under: /product/dsp/worker-config
- Add a pipeline/ directory
- Add replay-agent-config.json
Create a master branch (copied from main)
Current version requires master branch.
1.4 CMS Refresh
Restart CMS to apply agent changes:
docker-compose -f deploy-aws-25.3.2.yml -p dsp down
docker-compose -f deploy-aws-25.3.2.yml -p dsp up -d
docker restart <CONTAINER_IDS>
Start Replay Agent:
docker compose -f replay-agent.yml up -d
1.5 Pipeline Creation
Create a replay pipeline in CMS.
Example:
https://int-dsp-demo.krontech.com/pipeline/23246deb-5474-49ae-977c-4265dd6a5b58
1.6 S3 Destination Setup
Create an S3 bucket destination using environment details from:
https://int-dsp-demo.krontech.com/component/67150438-ce33-4e5b-9ace-0a02af6afb8b/
1.7 Replay Execution
- Navigate to Storage
- Select S3 bucket
- Choose a file (e.g. kagan-berk.log.gz)
- Click Replay
1.8 Monitoring Replay
- Select the pipeline
- Start replay
- Monitor status from:
- Agent Status panel
- Replay agent container logs on the server