DongHyeonkaandClaude Opus 5
b2963105a8
The keycloak project ended with four open questions that design could not
settle. A two-VM lab was built to answer them by measurement, and this is
that material: 26 experiments, 125 raw command outputs, 22 browser captures.
Follows the import procedure in README.md.
source/ the originating repository verbatim — 78 documents, 28 SVGs,
8 manifests, plus .source-revision recording the commit
final/ the SSOT
document.md 729 lines written from the 29 experiment documents, not
concatenated: what was predicted, what was measured, and
where the measurement itself was wrong
evidence/raw 125 outputs, flattened to <experiment>__<file> because
the originals collided (01-baseline.txt appeared three
times) and the audit only globs the top level
evidence/meta one per raw file; command and exitCode are null and the
README says why rather than inventing them
evidence/browser 22 captures
assets/ three diagrams through techviz
.techviz/ their VizSpecs
A separate project rather than an addition to keycloak: the B-layer answers
that project's four questions, but the A, C and D layers are about cluster
failure, SSO and operations, and one document.md should hold one subject.
The four question records there can point here through 관계.
Recorded rather than papered over: only three of the 28 diagrams were
remade. The repository forbids hand-drawn SVG and forbids titles inside the
canvas; all 28 originals carry both, so converting them is redrawing, not
reformatting. They stay in source/ and the gap is written into the document.
verify-pipeline.py passes. audit-records.py reports no issues.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
3.4 KiB
kind, slug, title, topic, project, status, sourceRevision, rootTreeNode, evidenceCapturedOn, assets, evidence, source, module
| kind | slug | title | topic | project | status | sourceRevision | rootTreeNode | evidenceCapturedOn | assets | evidence | source | module | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CONCEPT | grpc-observability-c03 | allowlist가 기본 거절이고 거절 목록은 메시지를 위한 것이다 | observability-models | clean-architecture-backend-template | 게시 전 | 21234e38cdb9a926cbc92bb97a2aee2e4a7d2916 | concept:grpc-observability-c03 | 2026-09-01 |
|
|
|
grpc-observability |
allowlist가 기본 거절이고 거절 목록은 메시지를 위한 것이다
violations(Map) 의 판정 순서가 셋이고, 그중 FORBIDDEN_TAGS 분기는 판정을 바꾸지 않고 진단만 바꾼다. 두 번째 분기의 allowlist가 이미 그것들을 거절한다.
본문
violations(Map) 의 판정 순서가 셋이다.
if (FORBIDDEN_TAGS.contains(key)) → "its value space grows with traffic…"
if (!ALLOWED_TAGS.contains(key)) → "not on the bounded allowlist [...]"
if (UNBOUNDED_VALUE.matcher(value)) → "looks like an identifier or a credential"
클래스 javadoc 이 두 목록이 겹치는 이유를 적는다 — "Everything unlisted is refused anyway; naming the dangerous ones gives the refusal a message that says why rather than just that." 즉 FORBIDDEN_TAGS 는 판정을 바꾸지 않고 진단만 바꾼다. 두 번째 분기가 이미 그것들을 거절한다.
허용 태그 여덟과 명시적 거절 열하나
허용 태그: grpc.service · grpc.method · grpc.rpc_type · grpc.status · grpc.channel_profile · grpc.completion_outcome · grpc.retry_bucket · grpc.stream_termination_reason.
명시적 거절: actor_id · tenant_id · object_id · stream_id · idempotency_key · request · response · metadata · authorization · error_detail · trace_id.
GrpcRpcObservation 참조 위치
:::evidence key="grpc-observability-c03" alt="코드베이스에서 GrpcRpcObservation 를 검색한 출력 5줄. 이 기록이 세는 참조가 그 출력에 그대로 보인다." caption="GrpcRpcObservation 코드베이스 검색 — 5줄 · exit 0" zoom="true" :::
값 검사가 잡지 못하는 것
UUID · sha256: 접두 · bearer 접두. 숫자 id, 이메일, 호스트명은 잡히지 않는다. 그리고 . 은 기본적으로 개행에 맞지 않으므로 값에 개행이 섞이면 matches() 가 거짓이 된다.
시도 수를 값이 아니라 버킷으로 접는다
retryBucket(int) 이 1-based 시도 수를 받아 0/1/2/3+ 로 접는다. 0 이하는 던진다. javadoc 이 이유를 적는다 — "an attempt count is unbounded in principle and the distinction anyone acts on is first attempt, one retry, several."
재시도 한 번이 호출 하나로 남는 이유
GrpcRpcObservation javadoc 이 적는다.
"A retried call is one observation with a retry bucket, and three attempt events beneath it; recording three separate calls instead makes the success rate read as 33% when the caller in fact got its answer."
그 분리가 GrpcObservationConvention.record(GrpcRpcObservation) 에서 실제로 그렇게 구현되어 있다 — RPC_DURATION 타이머는 1회, RPC_ATTEMPTS 카운터는 attempts 만큼 증가. 같은 태그 집합을 쓴다.