Record production launch-path verification

This commit is contained in:
SimpleTest 2026-08-09 09:42:43 +03:00
parent 029373aa80
commit c97744a777

View File

@ -51,6 +51,51 @@ revision and the two core production browser flows. They do not by themselves
prove external authentication-email receipt, browser Web Push delivery, or
production support/legal queue routing.
## Production authentication, queue routing, and operations recheck on 2026-08-09
- A real public passwordless-login request for the existing production account
`simpletestxxx@gmail.com` returned HTTP 302 to the login check-email page and
retained one hashed login token with the configured 15-minute lifetime. In
this code path the reserved token is deleted when SMTP delivery returns an
error, so the retained token proves that the configured production SMTP
adapter accepted this dispatch. It does not by itself prove arrival in the
external Gmail mailbox; mailbox receipt remains a separate observation.
- One run-scoped authenticated support case and one run-scoped authenticated
general content-removal notice were created for the existing owner account.
Both received `contact_verified_at` immediately, the support case appeared in
the permission-scoped support queue, and the notice appeared in the
permission-scoped legal queue. The removal acknowledgement was accepted by
the configured SMTP adapter. Exact-ID cleanup then deleted only those two
records and their audit events. Before/after totals returned to zero support
cases, zero removal notices, and one pre-existing audit event.
- Production configuration inspection observed operator-email mode `disabled`.
Consequently new support/legal intake is handled through the in-application
staff queues without sending an operator email for every case; requester
acknowledgements and staff replies remain transactional email paths.
- The daily encrypted off-site backup timer was loaded, enabled, active, and
waiting. Its latest completed service exited successfully after a PostgreSQL
18.4 custom-format dump, catalog and checksum verification, encrypted
off-server storage, and an isolated restore. The retained Restic snapshot is
`869e3e3a7444d726daef7e9afbdcc9ee082962a6121ec2e29c942600396fd13c`;
evidence is under
`output/production-operations/20260808T210010Z-2802200/`.
- The independent BuyVM monitor timer was active and waiting. Repeated latest
service executions exited successfully and recorded readiness HTTP 200 with
the expected ready payload plus a successful aggregate-metrics scrape. This
proves the current external probe execution, not an availability SLO.
- The active pilot rate-limit policies observed in production were: registration
and magic-link email 4/hour per address and 120/hour per IP; password login
10/15 minutes per address and 300/15 minutes per IP; email change 3/day per
account and 60/hour per IP; support intake 5/day per account and 120/hour per
IP; content-removal intake 20/day per account and 120/hour per IP. Four live
bucket rows were present at the inspection instant. These values are the
configured pilot policies backed by this project's earlier load checks, not
a universal capacity recommendation.
- After the checks, the compact production application remained healthy with
zero restarts and no OOM event, and readiness again returned the exact ready
response. The frozen hackathon-test checkout and containers were not changed,
and the public remote repository was not updated.
## Local support/legal and complete browser E2E on 2026-08-09
- The focused isolated Chromium run exercised an authenticated support case,