Yunxiao Flow Pipeline Troubleshooting

Scenario Description

This skill provides comprehensive troubleshooting and analysis capabilities for Yunxiao Flow pipeline failures. It helps diagnose issues across various pipeline stages, including:

Covered Pipeline Types:

• Image Build (Docker)
• Docker Deployment
• Kubernetes Deployment
• VM Deployment
• Function Compute Deployment
• Unit Testing (Python, Maven/Java, Node.js, Golang, PHP)

Architecture: Python Scripts → Yunxiao Flow API → Pipeline Execution Logs → Failure Analysis → Solution Recommendations.

Installation

The websocket-client library is required.

Environment Variables

Authentication

Before executing any commands, ensure you have a valid Yunxiao Personal Access Token. Token priority:

1. Command-line argument --token
2. Environment variable YUNXIAO_ACCESS_TOKEN

# Method 1: Configure via environment variable (recommended)
export YUNXIAO_ACCESS_TOKEN="pt-xxxxxxxxxxxxx"

# Method 2: Pass via script command-line argument
# --token pt-xxxxxxxxxxxxx

Security Rules:

• NEVER read, echo, or print the full token value (scripts have built-in masking, showing only the first 6 and last 6 characters)
• NEVER ask the user to input the token directly in the conversation
• NEVER hardcode token values in scripts
• Tokens should be obtained from Yunxiao Console > Personal Settings > Personal Access Tokens

RAM Permissions

This skill uses token-based authentication for Yunxiao Flow API. Required permissions:

• Flow Pipeline all modules: Read-only access

Parameter Confirmation

Important: Parameter Confirmation — Before executing any command or API call, all customizable parameters must be confirmed with the user. Never assume or use default values without explicit user approval.

Core Workflow

Step 1: Get Pipeline Run Status

Note: The --token \x3Ctoken> in the following commands can be provided via environment variable YUNXIAO_ACCESS_TOKEN. If the environment variable is set, the --token argument can be omitted.

python3 scripts/yunxiao_flow_get_pipeline_runs.py \
  --org-id \x3Corg-id> --pipeline-id \x3Cpipeline-id> \
  --pipeline-runid \x3Cpipeline-runid> --token \x3Ctoken> --domain \x3Cdomain>

Analyze the returned JSON:

1. Check the status field (SUCCESS/FAIL/RUNNING/INIT)
2. Traverse stages to find the failed stage
3. Find the failed job within the failed stage
4. Find the failed step within the failed job

Step 2: Get Failed Step Logs

python3 scripts/yunxiao_flow_get_job_step_log.py \
  --org-id \x3Corg-id> --pipeline-id \x3Cpipeline-id> --pipeline-runid \x3Cpipeline-runid> \
  --job-id \x3Cjob-id> --step-index \x3Cstep-index> --build-id \x3Cbuild-id> \
  --token \x3Ctoken> --domain \x3Cdomain> --full-log

Step 3: Get VM Deployment Logs (if applicable)

python3 scripts/yunxiao_flow_get_vm_deploy_machine_log.py \
  --org-id \x3Corg-id> --pipeline-id \x3Cpipeline-id> --pipeline-runid \x3Cpipeline-runid> \
  --deploy-order-id \x3Cdeploy-order-id> --machine-sn \x3Cmachine-sn> \
  --token \x3Ctoken> --domain \x3Cdomain>

Step 4: Consult Reference Documents for Analysis

Read the corresponding reference documents based on the failure scenario. For detailed guidance, see references/troubleshooting-guide.md.

Must read: yunxiao_base.md, yunxiao_flow_get_pipeline_runs.md

Step 5: Analyze and Provide Solutions

After confirming the build cluster and code clone information, output:

1. Pipeline basic information, which must include: build cluster (build cluster name must NOT be obtained from params.buildNodeGroup. It MUST be obtained from the "Request Runtime Environment" step logs.), build environment, trigger time, status, failed job, failed step, downloaded code source, job workspace (can be viewed from the PROJECT_DIR in the Pipeline Cache step)
2. Failed step information
3. Error log summary
4. Root cause analysis
5. Solution recommendations (configuration changes, parameter adjustments, operational steps)

Success Verification Method

For verification steps, see references/verification-method.md

Cleanup

No resource cleanup required. This skill only performs read operations for analysis purposes.

Best Practices

1. Always validate org-id length to determine organization type (24 chars = central org, 32 chars = regional org)
2. Check build cluster type from "Request Runtime Environment" step logs, not from params.buildNodeGroup
3. Read base reference documents first: Read yunxiao_base.md and yunxiao_flow_get_pipeline_runs.md before scenario-specific files
4. Extract build cluster information from log patterns:
   • Default: [Build Cluster Info] >> Cluster Name : Yunxiao China Hong Kong Build Cluster
   • VPC: [Build Cluster Info] >> Cluster Name : Hangzhou VPC Build Cluster
   • Private: [Runner Group Info]: >> RunnerGroupName : xxxxx
5. Verify code clone information: Check branch name and commit SHA before analysis
6. Match failure scenarios to corresponding reference documents in the troubleshooting guide table

Reference Document Links

Tools