G2 S6: Mac-tail cluster re-capture (live kind stand)#164
Merged
Conversation
…substitution documented)
…tion), not fabricated
MaterializedPostgreSQL self-bootstraps its publication with CREATE PUBLICATION executed as the replication user, and Postgres requires CREATE on the database for that. With CONNECT alone the replication handler retry-loops on "permission denied for database ops" — the CH database appears but carries no slot, no publication and no tables. Found live on the hq-demo kind stand 2026-07-06; the fan-out variant (fanout/03_cdc_setup.sql) always carried the equivalent grant. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…st ambient-noise caveat)
…-branch fan-out) Both sections now carry live output from the hq-demo kind stand (2026-07-06). §14: snapshot through the WAL pipeline (50/200/20/80), slot ops active with advancing flush LSN, publication with 4 tables, promotion collapsing to the same pg_ops__* hub counts as the §10 pull pass, live INSERT/UPDATE/DELETE round-trip, count()-vs-column-read tombstone semantics note, and the two-pass story (contention first, then the deterministic GRANT CREATE root cause fixed in cdc_setup.sql). §15: two per-branch Postgres DBs, two auto-named publications and slots, 10/30 + 8/20 snapshots, 12-second edit propagation with column-level detail, and a 0/0 cross-branch isolation check. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ion race Commit 1784d5e (B2-mp4 re-record, prepared in a parallel working session) was cut from a pre-2ed9ea3 copy of demo_evidence.md and unintentionally reverted §14/§15 to the earlier attempted-blocked account, and stamped the header with "§14/15 did not complete". Re-apply the full live capture from 2ed9ea3 verbatim and reword the header note to match reality (blocked first pass, then GRANT CREATE root cause found and the sections captured live). No other B2 content from 1784d5e is touched. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…nestly blocked by host contention
DORA Metrics
MTTR note: No failed mainline CI runs in the selected window. |
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.
Summary
demo_voiced.mp4/demo.cast/demo_transcript.txtlive on the current kitchen-legend narration (fixed two real bugs found along the way: an unconfiguredmc aliasinsideminio-0, and aReplacingMergeTreedouble-count from a retried promotion).postgres:16container,APPLY=1 verify_bv_order.sh, 17/17 PASS — transcript atdocs/perf/bv-order-canonical-pg-smoke-2026-07-06.md.demo_evidence.md§1-3, §9, §10, §12, §13 on a rebuilthq-democluster against the current seed. Two explicit, stated substitutions (§12 hand-ran the DAG's commands instead of installing Argo Workflows; §13 used direct SQL instead of adbtJob after it OOM'd) — not disguised as full runs.cdc_setup.sqlonly grantedrep_userCONNECT, butMaterializedPostgreSQLself-bootstraps its publication viaCREATE PUBLICATIONas that user, which needsCREATEon the DB. Both sections now carry a genuine live capture.load-test-baseline.md(12 scenarios, ~25 min under heavy host contention — explicit ambient-noise caveat included so the numbers aren't mistaken for an engine regression).helm lintclean,helm template --set replicaCount=2correctly fails without both ADR 0010 gate halves, succeeds with both). The live 2-pod deployment could not complete — a fresh app image build stalled twice under severe, sustained shared-host contention (co-tenant container at up to 80% CPU, this session's own kind control-plane crash-looping). Documented honestly indocs/clickhouse-cutover-plan.mdPhase 3 and ADR 0010 rather than faked.All of this ran live against a real Mac kind stand via SSH (host shared with another project's containers, which were left untouched). Every section states its own capture method; nothing here is fabricated or backfilled.
Test plan
🤖 Generated with Claude Code