fix(supabase): db:trial second re-review fixes

Post-create scan covers pg_stat_statements and labels each count; log path defined; rotate-secrets 401 window documented; plan shows image state unknown without a daemon; start_containers checks the app host before recreating GoTrue; truthful restore and destroy way-out messages; harness checks read stdin from /dev/null. Harness 372/0.
This commit is contained in:
bchanot committed 2026-11-11 14:36:12 +01:00
1 parent d3bdbbe0ba
commit f2cbf6814f
4 files changed
+121 -22

No files matched your search

@@ -3603,6 +3603,18 @@ Measured on Docker 29.9.3: a failing `inspect -f` (absent resource or lost daemo
- M7: `cmd_start` checks `trial_host` before starting or recreating a container.
- H: `lines_before_first` checks order; the rotation check asserts every silencing line; the tautological `! grep "already exists"` replaced by one that can fail; cases: stream split of the stub, plan, create, backup and status on stub-absent resources, create with the daemon lost after the first resource, restore with the daemon lost, start with no usable host.
#### Closure second re-review fixes (2026-11-11)
- I-1: SUPABASE.md §5.2 post-create scan also counts `extensions.pg_stat_statements` per key; above 0: reset it as `supabase_admin`, then `rotate-secrets`, existing backups keep the text.
- m1: `check()` runs every check with stdin from `/dev/null`; the two `grep ... $(find ...)` checks (rotation SQL, account body) get `/dev/null` as an extra file.
- m2/m3: `refuse_inspect` states a reason alone (`container_running`'s two refusals); `inspect_refused` hint adds the Docker 29.9.3 wording note; SUPABASE.md records it.
- m4: `plan_images` prints "unknown (Docker daemon unreachable)" when the daemon is down.
- m5: `start_containers` calls `trial_host` before the first start when GoTrue's container is missing.
- m6: harness cases: the quoted answer and hints of `inspect_refused`, the "listing failed" way-out lines (stub hook `lost-after-archive`).
- m7/m8: SUPABASE.md: stale access token keeps getting 401 after rotation until refresh or sign-in; log scan defines `<log_path>`, labels each count, states both log-location cases.
- m9: `restore_extracted_note` names what was extracted when a volume is not empty or unreadable (stub `volume-ls-storage`).
- m10: `destroy_stop_line` prints `docker stop <running>` before the `docker rm` way-out line.
## Review record (orchestrator)
Each reviewer was a fresh subagent; each fix round went to a fresh executor; the orchestrator re-ran the checks