Scaffold: Job Deployment

This document is an entrypoint for job deployment scaffolding. The step-by-step flow is in /.system/skills/ssot/repo/scaffolding/devops-deploy-job/SKILL.md.

1. Purpose & Scope

Purpose: Generate deployment skeleton for running a batch job in target environments (Kubernetes CronJob, scheduled tasks, etc.).

Scope:

•Creates job deployment configuration files
•Sets up schedule configuration
•Generates deployment scripts
•Records deployment decisions in workdocs

Out of scope:

•Actual deployment execution (requires human with credentials)
•Packaging/Docker build (see devops_packaging_job.md)
•HTTP service deployment (see devops_deploy_service.md)
•Infrastructure provisioning

2. Inputs & Preconditions

Required Inputs

Parameter	Type	Description
`job_name`	string	Job identifier (matches packaging)
`environment`	string	Target environment: `dev`, `staging`, `prod`

Optional Inputs

Parameter	Type	Default	Description
`image_tag`	string	`latest`	Docker image tag
`schedule`	string	""	Cron schedule (e.g., `0 2 * * *`)
`resources`	object	{}	CPU/memory limits
`env_vars`	object	{}	Environment variables
`secrets`	array	[]	Secret references
`concurrency_policy`	string	`Forbid`	CronJob concurrency: `Allow`, `Forbid`, `Replace`
`restart_policy`	string	`OnFailure`	Pod restart policy

Preconditions

•Job has been packaged (see devops_packaging_job.md)
•/ops/deploy/ directory exists
•Human has deployment credentials for target environment
•Target environment infrastructure exists

3. Step-by-Step Flow (AI + Human)

See /.system/skills/ssot/repo/scaffolding/devops-deploy-job/SKILL.md.

4. Tools & Scripts

Tool	Purpose
`scripts/devops/scaffold/devops_deploy_job.py`	Orchestrator script
`/ops/deploy/scripts/deploy.sh`	Generated deploy script

5. Outputs & Side Effects

Files Created

code

ops/deploy/jobs/<job_name>/<environment>/
  cronjob.yaml        # Kubernetes CronJob
  configmap.yaml      # ConfigMap (if env_vars)
  config.yaml         # Deployment configuration
  deploy.sh           # Deployment script
  run-once.sh         # Manual trigger script
  README.md           # Human-facing documentation

Kubernetes CronJob Template

yaml

# Auto-generated by devops_deploy_job scaffold
apiVersion: batch/v1
kind: CronJob
metadata:
  name: <job_name>
  namespace: <environment>
spec:
  schedule: "<schedule>"
  concurrencyPolicy: Forbid
  successfulJobsHistoryLimit: 3
  failedJobsHistoryLimit: 3
  jobTemplate:
    spec:
      backoffLimit: 3
      activeDeadlineSeconds: 3600
      template:
        spec:
          restartPolicy: OnFailure
          containers:
            - name: <job_name>
              image: <registry>/<job_name>:<image_tag>
              resources:
                limits:
                  cpu: "1000m"
                  memory: "1Gi"
                requests:
                  cpu: "100m"
                  memory: "256Mi"

Configuration File

yaml

job_name: <job_name>
environment: <environment>
image_tag: latest
schedule: "0 2 * * *"
concurrency_policy: Forbid
restart_policy: OnFailure
timeout: 3600
backoff_limit: 3
resources:
  limits:
    cpu: "1000m"
    memory: "1Gi"
  requests:
    cpu: "100m"
    memory: "256Mi"
env_vars: {}
secrets: []

6. Safety & Rollback

Safety Measures

•--dry-run always available for preview
•Human approval required before execution
•No actual deployment during scaffold
•Credentials not stored in generated files
•Production jobs require additional approval

Rollback Procedure

For job suspension (human executes):

bash

kubectl patch cronjob/<job_name> -n <environment> -p '{"spec":{"suspend":true}}'

For scaffold rollback:

•Delete the deployment directory:

bash

rm -rf ops/deploy/jobs/<job_name>/<environment>/

7. Job Operations (Human Required)

After scaffold, human manages job:

Deploy CronJob

bash

kubectl apply -f ops/deploy/jobs/<job_name>/<environment>/

Trigger Manual Run

bash

kubectl create job --from=cronjob/<job_name> <job_name>-manual-$(date +%s) -n <environment>

Check Job Status

bash

kubectl get cronjobs -n <environment>
kubectl get jobs -n <environment> -l app=<job_name>
kubectl logs -n <environment> job/<job_name>-xxx

Suspend/Resume Job

bash

# Suspend
kubectl patch cronjob/<job_name> -n <environment> -p '{"spec":{"suspend":true}}'

# Resume
kubectl patch cronjob/<job_name> -n <environment> -p '{"spec":{"suspend":false}}'

8. Related Documents

•/.system/skills/ssot/repo/scaffolding/devops-deploy-job/templates/deploy_template.md - Deployment knowledge template
•/.system/skills/ssot/repo/scaffolding/devops-packaging-job/SKILL.md - Job packaging scaffold
•/.system/skills/ssot/repo/scaffolding/devops-deploy-service/SKILL.md - Service deployment scaffold
•/ops/deploy/AGENTS.md - Deployment strategy