ixzc opened a new pull request, #3: URL: https://github.com/apache/doris-skills/pull/3
## Summary Add a comprehensive diagnostic skill suite for Apache Doris production troubleshooting, covering **10 diagnostic domains** with **16 detailed case files**, **45 production case patterns**, and structured reference documentation. ## What's included ### Skills (11 new) | Skill | Domain | |-------|--------| | `doris-debug` | Diagnostic router — symptom-to-skill mapping | | `doris-debug-query` | Slow/hanging/timeout queries, Profile analysis, Exchange/brpc | | `doris-debug-import` | Stream/Broker/Routine Load, Group Commit WAL | | `doris-debug-compaction` | -235 errors, version accumulation, base/cumu score | | `doris-debug-node` | OOM, crash, memory leak, BDBJE corruption | | `doris-debug-materialized-view` | Sync MV miss, MTMV refresh/rewrite | | `doris-debug-tablet` | Tablet/replica health, clone backlog, disk skew | | `doris-debug-deployment` | FE/BE startup, ports, priority_networks | | `doris-debug-data-lake` | Hive/Iceberg/Paimon/Glue catalog issues | | `doris-debug-resource-isolation` | Workload Group, spill, cgroup | | `doris-debug-cloud` | Storage-compute separation, file cache, meta-service | ### Case files (16) Each case follows a uniform format: **Environment → Symptom → Key Evidence → Investigation (step-by-step) → Root Cause → Fix → Key Diagnostic Actions**. Examples: - Plan Time 10min — Nereids Iceberg Create Scan Range RPC explosion - E11 Exchange brpc fanout — peak-hour fragment concurrency saturation - Import RPC timed out — BE Heavy Work Pool exhaustion - FileCache LRU Recorder queue pile-up → memory leak - Warmup shows FINISHED but zero data pulled — FE sends empty batch - Workload Policy ConcurrentModificationException — HashMap concurrent modification ### Shared references (5 + 2 guides) | Document | Content | |----------|---------| | `01-common-commands.md` | Diagnostic SQL, HTTP API commands | | `02-source-map.md` | Symptom → Doris source code mapping (with version annotations + upgrade caveats) | | `03-case-index.md` | 45 production case summaries across 8 domains | | `04-metrics-guide.md` | Per-domain key metrics with alert thresholds | | `05-config-quick-ref.md` | Per-domain key configuration parameters with tuning guidance | | `guides/` | Cross-module cascade troubleshooting (import→WAL→versions, query→exchange) | ### Companion skill integration - `doris-debug-query` routes to `doris-profile-reader` for deep Profile analysis (Cause C — BE compute bottleneck) - Works alongside existing `doris-best-practices` and `doris-architecture-advisor` ## Structure Each skill follows the Agent Skills format: ``` skills/doris-debug/SKILL.md # Router skills/doris-debug-query/SKILL.md # Cause table + Metric Taxonomy + Decision tree skills/doris-debug-query/references/ # case-*.md files ... skills/doris-debug-shared/references/ # Shared across all debug skills ``` ## Methodology 1. Client → FE → BE → Disk/Network top-down 2. Logs / Metrics / Profile / SHOW PROC evidence first 3. Stop the bleeding before root cause 4. Verify against Doris source code (source-map) 5. Match known case patterns to avoid rediscovery -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
