# Nightly Test Summary — run 2026-07-17→18 (IST)

**17th consecutive night of the "server/DB/proxy unreachable" outage plateau — no new regressions, no resolutions. 742 failed / 826 passed, both known persistent issues (tour_pg 502, rentveh stale SW_VER) still present.**

## Run identification
- Marker: `test_reports/nightly_marker.json` → `date: "20260717"`, `exit_code: 1`, `finished_at: 2026-07-17T23:23:52Z` = **2026-07-18 04:53:52 IST** (marker is UTC-run-start-stamped, this run belongs to today, 2026-07-18 IST — consistent with every prior night's naming quirk, not a stale/missing marker).
- Source: `test_reports/nightly_20260717.json`

## Totals
| | Count |
|---|---|
| Total | 1708 |
| Passed (expected) | 826 |
| Failed (unexpected) | 742 |
| Skipped | 140 |
| Flaky | 0 |

Prior night (2026-07-16→17): 1706 total, 810 passed, 756 failed, 140 skipped, 0 flaky.
Delta: passed +16, failed −14, skipped unchanged. Total +2.

## Regressions vs baseline (2026-06-20: 0 failed) and vs the ongoing 16-night "server/DB/proxy unreachable" plateau

This is the **17th consecutive night** of the mass-failure pattern (timeouts / 404 raw Apache HTML for manifest.json / 406-400 PostgREST `Accept-Profile` errors) affecting the same subset of PWAs. Per-file failure counts match the recent plateau for the worst offenders: fundcustodian 154, sales_billing 78, vendors 74, upi_pay 68, customers 50, custodian 40, recce_client 32, recce 32, sales 32, safe_bottom 28, stores 26, counters 18, yagya-portfolio 18, contacts 14, finance/installation/tour_planner 10 each, hub/oc_admin/vrs 8 each, admin/smoke 6 each, client 4, debug_tour/hr/rentveh/tour_pg 2 each. Green set unchanged (printing, activity, drive_consolidator, hr-app pages, expense, dispatch, production, vehicle, btl, tour_pg_sharing). The +16 passed / +2 total vs. last night is consistent with new tests being added and passing, not with the outage improving.

**STILL PRESENT — rentveh.spec.js `SW cache version matches CACHE_VER in app` (android-chrome + desktop-chrome, 2 instances):** unchanged from last night. Test still asserts `sw.js` contains `'rentveh-v3'`; actual `sw.js` ships `CACHE = 'rentveh-v5'`. Stale hardcoded `SW_VER` constant in the spec, not an app defect — still needs `tests/rentveh.spec.js` bumped to v5.

**STILL PRESENT — tour_pg.spec.js `matrix + 2-opt path for 14 stops` (android-chrome + desktop-chrome):** still failing (502 from tour-pg-proxy), 5th night running as a durable defect, unrelated to the server-unreachable plateau. Needs tour-pg-proxy investigation.

## Baseline
`project_pwa_test_coverage.md` updated with tonight's numbers below — still not re-baselined. The 2026-06-20 0-failed baseline remains the trusted reference until the dev-stack uptime issue and the tour_pg 502 are fixed and a clean run comes back.