| name | spark-operations-cli |
| description | Diagnose failed Spark jobs, unhealthy Livy sessions, and performance bottlenecks in Microsoft Fabric via read-only CLI triage. Use when the user wants to: (1) diagnose why a Spark job, notebook run, or Lakehouse job failed, (2) triage stuck or dead Livy sessions, (3) identify OOM, shuffle spill, or data skew, (4) retrieve driver and executor logs or Spark Advisor findings, (5) copy event logs and start a local Spark History Server, (6) diagnose all Spark activities within a failed pipeline run. Triggers: "diagnose my failed notebook", "why did my spark job fail", "triage spark failure", "diagnose pipeline run failure", "why did my pipeline fail", "livy session stuck in starting", "spark executor OOM", "check spark advisor findings", "shuffle spill diagnosis", "why did my lakehouse job fail", "diagnose lakehouse table load", "data skew diagnosis", "open spark history server locally", "analyze spark failure logs", "spark job triage".
|
Update Check — ONCE PER SESSION (mandatory)
The first time this skill is used in a session, run the check-updates skill before proceeding.
- GitHub Copilot CLI / VS Code: invoke the
check-updates skill.
- Claude Code / Cowork / Cursor / Windsurf / Codex: compare local vs remote package.json version.
- Skip if the check was already performed earlier in this session.
CRITICAL NOTES
- To find the workspace details (including its ID) from workspace name: list all workspaces and, then, use JMESPath filtering
- To find the item details (including its ID) from workspace ID, item type, and item name: list all items of that type in that workspace and, then, use JMESPath filtering
- Skill disambiguation:
spark-operations-cli is for read-only triage and diagnosis of existing jobs and sessions. For creating notebooks, running new jobs, or Spark development, use spark-authoring-cli. For interactive PySpark analysis and Livy session creation, use spark-consumption-cli.
Spark Operations — CLI Skill
This skill provides diagnostics for Microsoft Fabric Spark job failures, Livy session health, and performance bottlenecks using Fabric REST APIs and CLI tools (az rest). All diagnostic operations are read-only; session cleanup (e.g., stopping zombie sessions) requires explicit user confirmation. For Spark development and notebook authoring, use spark-authoring-cli. For interactive PySpark analysis, use spark-consumption-cli.
Table of Contents
The TOC is grouped by purpose. Start at Diagnostic Workflows when triaging an active failure; the earlier sections are foundational references.
1. Fabric Foundations (concepts)
2. CLI Setup & Authentication
3. Spark Sessions, Notebooks & Jobs (background)
4. Spark Monitoring APIs (primary triage surface)
5. Diagnostic Workflows (start here for active triage)
6. Job Failure Diagnostics
7. Livy Session Health
8. Performance Diagnostics
9. Offline / Deep-Dive Tools
Must/Prefer/Avoid
MUST DO
- Always retrieve job/session status before attempting remediation
- Use workspace and item discovery from COMMON-CLI.md — never hardcode IDs
- Check Livy session state before submitting diagnostic statements
- Follow the Failure Triage Workflow for systematic diagnosis
- Always check the Spark Advisor API before reading raw logs — it often identifies the root cause immediately
- Use monitoring APIs (no active session required) before attempting Livy-based diagnostics
- Poll job/session status with 10–30 second intervals; timeout diagnostics after 30 minutes
- Always include the Notebook Snapshot URL in diagnostic output — it has the longest retention and enables cell-level inspection in the Fabric UI
PREFER
-
Querying job instance history to establish baseline before declaring a regression
-
Reusing existing idle sessions for diagnostic queries instead of creating new ones
-
Checking capacity utilization when jobs are slow before blaming the Spark code
-
Using az rest with JMESPath filtering to extract specific fields from large API responses
-
The Spark Advisor API over manual log parsing for skew, task errors, and timeout detection
-
Resource Usage API coreEfficiency metric to quantify cluster utilization before recommending scaling
-
Job instance history comparison (last 5 runs) to detect regressions before deep-diving
-
For MLV refresh scheduling, monitoring, or run-history, use mlv-operations-cli. For diagnosing the underlying Spark job failure (OOM, skew, shuffle spill), continue with this skill — MLV refreshes execute as Spark jobs and their logs are accessible via the same monitoring APIs.
-
MLV failure classification — when diagnosing a failed MLV refresh, classify the error before deep-diving:
| Error Pattern | Category | Diagnosis Path |
|---|
MLV_SPARK_SESSION_REQUEST_SUBMISSION_FAILED | Infrastructure | Capacity paused/unavailable, Spark pool misconfigured. Check capacity state first. |
MLV_SELECTED_NOT_FOUND | Configuration | MLV table was deleted/renamed. Verify table exists via SHOW MATERIALIZED LAKE VIEWS IN schema. |
OutOfMemoryError / SparkOutOfMemory | Resource | Source data grew beyond cluster capacity. Check Spark Advisor for memory pressure. |
ShuffleBlockFetchFailed / data skew | Performance | Uneven data distribution. Use Resource Usage API to identify skewed partitions. |
DeltaTableVersionNotFound | Dependency | Source table was vacuumed below retention threshold. Extend delta.logRetentionDuration. |
ConstraintViolationException / ON MISMATCH | Data Quality | DQ constraint dropped/failed rows. Check source data quality upstream. |
| Timeout (run > 24 hours) | Scale | Lineage too large for single run. Split into smaller lineage groups across lakehouses. |
AVOID
- Killing sessions without checking if they have active statements
- Creating new sessions for every diagnostic query (reuse idle sessions)
- Assuming OOM without checking actual memory metrics from Livy
- Hardcoded workspace or item IDs in diagnostic scripts
- Diagnosing performance without first checking capacity throttling via the Admin API
- Submitting diagnostic statements to sessions in
busy state
Examples
Example 1: Diagnose a Failed Notebook
User prompt: "Why did my notebook ETL_Daily fail in workspace Production?"
Agent workflow:
- Resolves workspace →
workspaceId, item → itemId (Notebook)
- Lists recent Livy sessions, auto-picks the Failed session
- Queries Spark Advisor → finds
TaskError: OutOfMemoryError on executor
- Queries
/stages → confirms data skew (12× max/median ratio in stage 5)
- Presents report with HIGH findings + fix recommendations
Example 2: Triage Stuck Livy Session
User prompt: "My Livy session abc-1234 is stuck in starting state"
Agent workflow:
- Uses session ID directly, queries session state
- Lists all workspace sessions → detects 8 concurrent sessions (capacity pressure)
- Checks Livy log → no errors, just queued
- Reports: capacity contention, recommends waiting or cancelling idle sessions
Example 3: Pipeline Failure Root Cause
User prompt: "Diagnose pipeline run 5678 in workspace Analytics"
Agent workflow:
- Resolves pipeline, calls
queryActivityRuns for run 5678
- Finds 2 Notebook activities: one Succeeded, one Failed
- Extracts
output.result.error.{ename, evalue, traceback} from failed activity
- Constructs Notebook Snapshot URL for cell-level inspection
- Presents error details + snapshot link + suggested fix
Quick Start
Environment Setup
Apply environment detection from COMMON-CLI.md to set:
$FABRIC_API_BASE and $FABRIC_RESOURCE_SCOPE$FABRIC_API_URL and $LIVY_API_PATH for Livy operations
Authentication: Use token acquisition from COMMON-CLI.md § Authentication Recipes.
Automated Diagnostic Workflow
When the user provides a simple prompt (e.g., "Diagnose my notebook ETL_Pipeline", "What's wrong with Spark application abc-123", "Check workspace Production for issues"), follow this fast-path summary. For full procedure, edge cases (expired data, pipeline-only sessions), report templates, and retention details, see references/automated-diagnostic-workflow.md.
Entry Points (what the user provides)
| User provides | Agent resolves |
|---|
| Workspace name | → workspaceId (via workspace list + name filter) |
| Notebook / SJD / Lakehouse name | → itemId (via item list + name/type filter) |
| Pipeline name + run ID | → child Spark activities → see pipeline-diagnosis.md |
| Livy session ID or Spark app ID | → Use directly |
| Nothing specific | → Ask for workspace name + item name |
Item-Type API Paths
| Item Type | Livy Sessions Path | Job Instances Path |
|---|
| Notebook | /notebooks/{id}/livySessions | /items/{id}/jobs/instances |
| Spark Job Definition | /sparkJobDefinitions/{id}/livySessions | /items/{id}/jobs/instances |
| Lakehouse | /lakehouses/{id}/livySessions | /lakehouses/{id}/jobs/instances |
All session API paths follow: $FABRIC_API_URL/workspaces/$workspaceId/<itemTypePath>/$itemId/livySessions/$livyId/applications/$appId/<endpoint> — see SPARK-MONITORING-CORE.md.
Steps at a Glance
| Step | When | Action | Auto-flag rule |
|---|
| 1. Resolve & Discover | Always | Resolve workspace → item → list recent Livy sessions; auto-pick if unambiguous, else prompt user | — |
| 1b. Fallback | Session 404 / Spark Monitoring data expired | Try queryActivityRuns (pipeline) → Job Instance failureReason → construct Notebook Snapshot URL | See reference § Step 1b |
| 2. Route by state | After Step 1 | Failed → 3+4+5 · Succeeded/InProgress → 4+5 · Cancelled → log+3 · idle/busy/starting → 6 · dead/killed/error → 3+6 | — |
| 3. Failure analysis | Failed / Cancelled / dead | Query in order: Spark Advisor → driver stderr → Job Instance → executor logs → Livy log → Resource Usage. Stop when root cause clear. | Match against job-diagnostics.md § Quick Reference Table |
| 4. Performance | Always (except 1b path) | /stages, /allexecutors | skew max/median > 3× · spill diskBytesSpilled > 0 · GC jvmGcTime/executorRunTime > 20% · shuffle > 1 GB · tasks < 100ms |
| 5. Resource utilization | Always (except 1b path) | /resourceUsage | coreEfficiency < 0.3 → HIGH · idleTime/duration > 0.4 → MEDIUM |
| 6. Session health | Idle/zombie checks | GET /workspaces/$workspaceId/spark/livySessions | idle + no recent statements → zombie · starting beyond expected → capacity |
| 7. Compile report | Final | Severity-ordered findings table + Notebook Snapshot link + suggested fixes | See reference § Step 7 for template |
Key principle: Always check Spark Advisor first — it's pre-computed and identifies most root causes without log parsing. Pipeline runs have the richest error data via queryActivityRuns (ename, evalue, traceback, cell/line) — see pipeline-diagnosis.md.
Data retention warning: Spark Monitoring API data (logs, stages, advisor) typically expires in minutes to hours after session end. Diagnose failures promptly. If APIs return 404, jump to Step 1b in the reference.
Tier 2 escalation: For truncated data, HTTP 408/504, or DAG/SQL plan visualization, suggest the offline Spark History Server workflow.
