admin/38686-vm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: production fully deployed at 1d224bc + capture deploy-ordering lesson

Browse Source 
Production caught up — all 14 pending commits live after a
second service restart (the first restart ran before the code
reached the target commit; DEBUG=False's cached template loader
held the old templates until restarted again).

CLAUDE.md:
- 'What's mid-flight' breadcrumb: no longer says pending deploy;
  now states production is at 1d224bc and fully live.
- Flatlogic Deployment section: new '⚠ DEPLOY ORDERING' bullet
  documenting that production runs DEBUG=False → cached template
  loader → restart MUST come after the pull, and template-only
  changes still need a restart (unlike DEBUG=True local dev).
  Includes the symptom ('git log shows right commit but page
  looks old') and the fix (restart again).
- Bumped the {# #} bit-us count + added a grep sanity-check
  one-liner (from the prior commit, retained).

parked-work.md: 'Pending pull-and-restart' section replaced with
'Production status — fully caught up'.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Konrad du Plessis 2026-05-15 11:29:26 +02:00
parent 1d224bc01b
commit 5162db966a
2 changed files with 25 additions and 43 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
CLAUDE.md
View File

@ -3,21 +3,14 @@
## What's mid-flight — read this first
**Parked / deferred work:** see `docs/plans/parked-work.md`.
**Production status (15 May 2026):** migrations `0013_add_site_report`,
`0014_add_absence`, `0015_absence_project` are deployed; `/history/`
is no longer crashing on the production VM. The Worker Absences
feature shipped on 14 May 2026 (commits `bf6f0a5` → `27fe05e` on
`ai-dev`). **13 subsequent commits** of UX polish + cleanup +
dashboard/report number audit (team filters, dropdown stacking fix,
day-name in payroll modal, calendar-month math, outstanding totals
aligned, `Project.name` dedupe, `timezone.localdate()`,
`_company_cost_velocity` SQL-aggregated, signed adjustment amounts,
hero-card scope sublines) are on `ai-dev` HEAD but not yet on
production — needs a `git pull` + `sudo systemctl restart
django-dev.service` whenever convenient (no migrations or
collectstatic required for any of them). See
`docs/plans/parked-work.md` "Pending pull-and-restart" table for
the full commit list.
**Production status (15 May 2026):** ✅ fully caught up. Production
is deployed at `origin/ai-dev` HEAD `1d224bc` — SiteReport +
Absences feature, all UX polish, team filters, the 18-finding
dashboard/report number audit, and template-comment fixes are all
LIVE. Nothing pending deploy. (Deploy-ordering lesson from this
day is captured in the "DEPLOY ORDERING" bullet of the
Flatlogic/AppWizzy Deployment section — restart must come AFTER
the code is pulled, because `DEBUG=False` caches templates.)
Phase A.2 (manual JournalEntry UI) and Phase B (Letterly inbound
webhook) from the Site Work Logging design are parked pending Q5 / Q7
@ -863,6 +856,7 @@ Gemini/shell — the user cannot modify via Flatlogic's web editor because
- **Migrations**: Sometimes run automatically during rebuild, but NOT always reliable. If you get "Unknown column" errors after pulling latest, visit `/run-migrate/` in the browser to apply pending migrations manually. This endpoint runs `python manage.py migrate` on the production MySQL database.
- **Static files**: Flatlogic's rebuild does NOT auto-run `collectstatic`. After CSS/JS changes have Gemini run `python3 manage.py collectstatic --noinput` + restart the service, otherwise Apache keeps serving the previously-collected copy.
- **Service**: The Django app runs as `django-dev.service` (systemd). Gemini restarts it via `sudo systemctl restart django-dev.service`. It runs `python manage.py runserver 0.0.0.0:8000` — a **development server**, not gunicorn/uwsgi (Flatlogic default, works fine at this scale).
- **⚠ DEPLOY ORDERING — pull THEN restart, not the reverse.** Production runs `DEBUG=False`, so Django uses the **cached template loader**: every `.html` template is compiled into memory once at process start and is NEVER re-read from disk until the process restarts. Symptom of getting this wrong: "I pulled the code, `git log` shows the right commit, but the page still looks old." Cause: the `restart` happened *before* the code reached the target commit (e.g. Flatlogic auto-pulled afterward, or Gemini pulled after restarting). **Fix: restart AGAIN, after confirming `git log --oneline -1` is at the target commit.** Correct deploy order is ALWAYS: (1) `git fetch github ai-dev && git reset --hard github/ai-dev`, (2) `/run-migrate/` if there are new migrations, (3) `collectstatic` if `static/` changed, (4) `sudo systemctl restart django-dev.service` **last**. Template-only changes still need the restart (cached loader) — unlike local dev where `DEBUG=True` re-reads templates per request. Bit us 15 May 2026: 14 commits of template fixes were "invisible" on prod until a second restart. `git reset --hard github/ai-dev` (not `git pull`) is preferred because the VM accumulates Flatlogic-editor autosave commits that make a plain pull conflict.
- **CDN**: All production traffic goes through Cloudflare. Response headers show `cf-ray`/`cf-cache-status`. Static assets are cached at the edge for 4h — see "Static Assets & Cache-Busting" section for how the `deployment_timestamp` token breaks stale caches.
- **Never edit `ai-dev` directly on GitHub** — Flatlogic pushes overwrite it
- **Gemini gotcha**: Flatlogic's Gemini AI reads `__pycache__/*.pyc` and gets confused. Tell it: "Do NOT read .pyc files. Only work with .py source files."

44
docs/plans/parked-work.md
View File

@ -7,36 +7,24 @@
---
## ⚠ Pending pull-and-restart on production
## Production status — ✅ fully caught up (15 May 2026)
**Status:** The big absences/SiteReport deploy (migrations
`0013`/`0014`/`0015` + collectstatic + service restart) was
completed on 14 May 2026 — `/history/` no longer crashes,
`/absences/*` is reachable, badge CSS is collected. ✓
Production (`https://foxlog.flatlogic.app/`) is deployed at
`origin/ai-dev` HEAD **`1d224bc`**. Everything is live: SiteReport
+ Absences feature, all UX polish, the team filters, the
dashboard/report number audit (18 findings), and the
template-comment + subline-clarity fixes.
Since that deploy, four small commits have shipped to `origin/ai-dev`
that are NOT yet on production:
| Commit | What it does |
|---|---|
| `4368e53` | Fixes the absence-form team filter (workers were all disappearing when a team was selected — read input.value instead of data-attr) |
| `02c6d4d` | Fixes the Reasons multi-checkbox dropdown stacking-context bug on `/absences/` (lifted the filter card above the table card) |
| `4b57cff` | Adds a Team filter to `/workers/` |
| `398a5b2` | Adds a Team filter to `/history/` (and CSV export) |
| `70fa085` | Shows day name in Work Log Payroll modal header ("Friday, 15 May 2026") |
| `d1d3e15` | 7 absence-feature polish follow-ups (dead code removal, N+1 fixes, DRY helpers, WCAG icon contrast) |
| `18c75b2` | Dashboard "Paid This Month" now actually uses calendar month (was 60-day rolling) |
| `2e6b78d` → `c02edce` | Dashboard + report numbers audit fixes — 18 findings, 6 commits: outstanding-total alignment, project-name dedupe, timezone.localdate(), off-by-one date windows, dead context removal, SQL-aggregate cost velocity, clearer labels, signed adjustment amounts, daily_rate documentation |
**To deploy these:**
- No migrations needed (pure template + view changes; no schema changes).
- No `collectstatic` needed (no new CSS or JS files).
- Just: ask Gemini in Flatlogic to `git pull origin ai-dev` and then `sudo systemctl restart django-dev.service`.
That's the whole sequence. Production should pick up all four
improvements instantly. Rollback is `git reset --hard <previous SHA>`
on the VM if anything looks wrong (no data migrations were involved,
so rollback is safe).
**Deploy-ordering lesson learned (now in CLAUDE.md):** production
runs `DEBUG=False` → Django's cached template loader holds compiled
templates in memory until the process restarts. The 14 commits of
template fixes were briefly "invisible" because the service was
restarted *before* the code reached the target commit. The fix was
simply to restart AGAIN once `git log` confirmed the right commit.
Correct order is always: `git reset --hard github/ai-dev` → migrate
(if needed) → collectstatic (if `static/` changed) → restart
**last**. See the "DEPLOY ORDERING" bullet in CLAUDE.md's
Flatlogic Deployment section.
---