Merged
Conversation
1. Fix docker-compose for OTLP traces e2e test: - Add health checks to frontend and productcatalogservice containers - Use depends_on with condition: service_healthy for proper startup ordering - Replace no.exist:80 endpoints with reachable addresses to avoid DNS/connection timeouts - Increase memory limits (200M->300M frontend, 20M->40M productcatalogservice) 2. Add e2e expectation specification documents covering all query protocols: - Core template syntax (contains, notEmpty, gt/ge/lt/le, b64enc, regexp) - GraphQL/MQE with .graphqls schema references - LogQL, PromQL, TraceQL, Zipkin v2, Status/Debug endpoints - CLAUDE.md guide for navigating e2e test structure
OTel demo images are Alpine-based (no bash, only sh). The bash /dev/tcp trick doesn't work. Use nc -nz (same as base-compose BanyanDB healthcheck pattern).
The CI failure was caused by the bash healthcheck, not memory. Restore original limits: 200M frontend, 20M productcatalogservice.
Remove 7 protocol-specific docs that duplicated info already in .graphqls schemas and existing expected files. Replace with a single CLAUDE.md focused on how to write accurate expectation files: - Exact function semantics (notEmpty, gt/ge/lt/le, regexp, b64enc) - contains action for unordered subset matching - Practical recipes: key existence, list min N records, nested objects, null handling, variable reuse, conditional rendering - Common pitfalls (type mismatches, ordering, whitespace)
The core policy: verify exact values for all meaningful fields (service names, labels, components, layers, scopes, tag keys/values), only use notEmpty/gt for genuinely dynamic values (timestamps, UUIDs, IPs). Updated all recipes to demonstrate accurate verification with real SkyWalking field patterns.
When writing expected files, Claude often cannot determine which fields are meaningful domain values vs dynamic runtime values. Added explicit workflow: show complete raw output, propose expected file with per-field reasoning, ask developer to confirm or correct.
wankai123
approved these changes
Mar 20, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fix OTLP traces e2e test instability
The OTLP traces e2e test has been flaky due to infrastructure issues (not sampling rate — that's 100% everywhere).
Root causes identified and fixed:
No health checks on OTel demo containers — trigger fired before services were ready, producing no traces during the retry window.
healthcheckwith TCP checks (same pattern as base-compose OAP)depends_on: condition: service_healthyfor proper startup orderingNon-existent service endpoints causing 20-30s timeouts —
CURRENCY_SERVICE_ADDR: no.exist:80andFEATURE_FLAG_GRPC_SERVICE_ADDR: no.exist:80caused DNS resolution failures and gRPC dial timeouts on every request, making/api/productsslow or failing entirely.productcatalogservice:3550(reachable endpoint, fast gRPC "unimplemented" error instead of hanging)Tight memory limits —
productcatalogserviceat 20M andfrontendat 200M could OOM under CI load.Also adds e2e expectation specification documents (CLAUDE.md and protocol-specific specs) for AI-assisted e2e test development.
Explain briefly why the bug exists and how to fix it.
Update the
CHANGESlog.