Why: The follow-up CI exposed that backend #188 broke the Dev
multi-instance suite: serve --mock seeded runtime.mock at the user
scope, which system-managed settings now reject, so instances never
became healthy. Backend #190 seeds the startup flag into the system
scope; this PR repins and drops the now-invalid alice user-scope write.
Evidence: Backend #190 CI success at head 0af1ebc (GATE); this PR's
composite CI re-runs make test-multi-instance (GATE for the fix).
Impact: backend pin 741da4a; multi-instance test comments updated to
the system-managed regime.
Refs: Backend #188, #190.
Why: include the current integration baseline before verification.
Evidence: fetched origin/main=3fb8210f6495a78e70a67af3661bdfe9c00fb9de.
Impact: integrate upstream changes without rewriting task history.
Refs: task-sync
Why: test-pg only cloned the default branch, so prev_sha and legacy_sha were never built or attested.
Evidence: SELF python3 -m unittest tests.test_ci_credentials tests.test_lifecycle_pin (11 tests, exit 0).
Impact: the lifecycle suite prints the combination inventory and keeps login and rdm.ping as NOT_RUN.
Refs: .task/feature-svc-d3-real-scenarios.md
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: A real process must not keep the caller DSN in its command line, environment, log, history, or svc.env. Domain events written by -bootstrap have to survive a later pin deploy.
Evidence: SELF unittest of the two new LifecycleRealTests methods exited 0.
Impact: The scan uses a sentinel that is not a live database password. Login and rdm.ping stay NOT_RUN.
Refs: tests/integration/lifecycle/test_lifecycle_real.py
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: A static dev deploy deleted the worktree before copying dist, so a mixed batch could not update the workbench. Missing manifests and a second instance on the same Postgres database must fail closed the same way on real artifacts.
Evidence: SELF unittest of the mixed batch, the three-entry manifest-loss case, and the PG cross-instance case exited 0. The PG case used a temporary cluster.
Impact: static dev deploys copy dist before recycling the worktree. Login and rdm.ping stay NOT_RUN.
Refs: tests/integration/lifecycle/test_lifecycle_real.py
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Stub journals do not prove that a real RDM binary rolls back after a switched fault, a failed build, or a candidate that exits on serve.
Evidence: SELF unittest of the three new LifecycleRealTests methods exited 0. Staging leaves the old pid and release in place; the next start clears that journal. Switched and an immediate serve exit both restore the previous real binary.
Impact: The crash candidate delegates version and identity to the real binary so the gate still sees a capable artifact. Login and rdm.ping stay NOT_RUN.
Refs: tests/integration/lifecycle/test_lifecycle_real.py
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: The lifecycle gate only ran four realcombo cases. A legacy candidate and a missing or mismatched identity row must be refused before the old process stops, and the prev/sha pins must switch without changing the identity row.
Evidence: SELF unittest of the four LifecycleRealTests methods exited 0 against a private RDM clone. The pin moved to the identity merge and a later commit on the same backend pin, because the previous card-close SHAs build a binary with no ensure/verify.
Impact: make test still skips this class unless SVC_TEST_REAL, SVC_RDM_SRC_DIR and SVC_REAL_RDM_BIN are set. make test-lifecycle now runs the module. Login and rdm.ping stay NOT_RUN.
Refs: tests/integration/lifecycle/rdm.pin
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: include the current integration baseline before verification.
Evidence: fetched origin/main=5aebedebeb963b5c01f39fd9122ec60dbe6d34d3.
Impact: integrate upstream changes without rewriting task history.
Refs: task-sync
Why: an unset object store must keep artifact bytes available to the other process, which the store-level test does not execute.
Evidence: cross_artifact_roundtrip failed with EOF on backend c6fb130 and passed on 46073d30 (SELF, isolated u7drill-k4).
Impact: Dev now builds the fixed backend. Second-instance write traffic stays closed.
Refs: modules/backend tests/integration/multi-instance/multi_instance_test.go
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: The live deploy journal never called fault injection, and test-lifecycle failed closed on login so CI could not require the real combination cases.
Evidence: SELF python3 -m unittest tests.test_svc_u2.U2ReleaseTests -q exit 0 (35 ok, 1 skipped) on 3602a87; receipt sha256 9b5785577751e8e9e7947cf9700fb8854b1c80142f34273ae25f3097ee4c9aa8.
Impact: test-pg builds the Dev binary and runs make test-lifecycle. Login and full-track deploy without SVC_RDM_ORIGIN stay NOT_RUN. No live data move and no account CLI.
Refs: recovery plan D2/D3; Dev #485
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: an update that stops the old process and overwrites the binary cannot
get the previous version back, and a migration the old binary cannot read
must be refused before that stop. CI also had no fixed RDM SHA to build from.
Evidence: python3 -m unittest tests.test_svc_identity tests.test_svc_recover tests.test_svc_deploy_txn tests.test_lifecycle_pin passed. make test-lifecycle exits 2 when AGENT_PG_TEST_DSN is unset.
Impact: installs land in releases/<rid> and current switches with mv -T.
Deploy keeps the desired state and rolls back a non-terminal journal.
Schema compatFloor above the installed target is refused unless
--migrate-schema --confirm is paired with a backup. test-lifecycle fails
closed without PostgreSQL and an RDM checkout. Login stays NOT_RUN.
Refs: recovery plan r3.1 D2 and D3; Backend feature/d1-identity-open
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Backend login_keys gate blocks Key traffic while
must_change_password is set; U7 failed on PUT settings/user 403.
Evidence: go test -c multi-instance compiles; CI test-pg will GATE.
Impact: loginAlice does session password change alicePass0→alicePass.
Refs: Dev PR #426; Backend #170 pin e7b70c2
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: collaboration storage U7 (proposal sections 3/5/7) requires two-process
HTTP-level acceptance over shared PG, a reproducible load-test entry honoring
the load contract, and an auditable pre-release checklist; capacity windows and
dual-host runs need user-scheduled resources, so the deliverable here is the
verifiable entry, not capacity claims.
Evidence: make test green (composite); make test-multi-instance green locally
against an isolated PG (five scenarios incl. crash-takeover and restart
invariants); run_load.sh --profile smoke end-to-end green (SSE probe p95~18ms,
zero seq duplicates, stack auto-recycled); backend pinned to feature branch
with the three cross-instance fixes.
Impact: new tests/integration/multi-instance (CI test-pg job step), new
tests/load/collaboration (k6 + SSE probe + run_load.sh), release checklist at
docs/release-multi-instance.md; no runtime behavior change in Dev repo itself.
Refs: backend PR #164; task card feature/u7-load-fault-release.
weichen
