feat(parse-server-mongo): compat-lane-ready sample + JS line coverage gate by AkashKumar7902 · Pull Request #97 · keploy/samples-typescript

AkashKumar7902 · 2026-05-01T01:19:42Z

Summary

Brings the parse-server-mongo/ sample (originally introduced as a focused boot-phase mongo tiebreaker reproducer in #94 / #95) up to the keploy compat-lane shape used by samples-python/doccano-django and samples-java/restheart-mongo:

flow.sh refactored into the bootstrap | record-traffic | coverage subcommand contract.
docker-compose.yml now uses ${VAR:-default} substitutions throughout so concurrent matrix cells can share a docker daemon.
New keploy.yml.template ships parse-server's noise filter (objectId, sessionToken, createdAt, updatedAt, Date header).
Broader REST + GraphQL traffic surface in parse_record_traffic (51 distinct fires across users, sessions, classes, roles, files, cloud-functions, schemas, hooks, GraphQL, aggregate).
Real JS line coverage of node_modules/parse-server/lib via NODE_V8_COVERAGE + a custom report tool — see "Coverage architecture" below.

The boot-phase divergence behaviour for the mongo/v2 tiebreaker test is preserved (bootstrap still sleeps 3s after /health so pre-mutation find _SCHEMA snapshots land first; record-traffic still issues the three-class _SCHEMA mutation pattern).

Layout

parse-server-mongo/
├── Dockerfile               # node:20-bookworm-slim + parse-server deps (base; uninstrumented)
├── Dockerfile.coverage      # extends base, adds c8 + graceful-shutdown shim
├── docker-compose.yml       # mongo:7 + this sample, env-driven
├── docker-compose.coverage.yml  # overlay; sets NODE_V8_COVERAGE=/coverage
├── coverage-report.js       # reads V8 dumps, emits Covered N/M (XX.X%)
├── flow.sh                  # bootstrap | record-traffic | coverage
├── index.js                 # 25-line parse-server bootstrap (unchanged)
├── package.json             # parse-server@8.2.3 + express@4.21.2 (unchanged)
└── keploy.yml.template      # noise filter

Coverage architecture

Real JS line coverage via Node-native NODE_V8_COVERAGE + a custom coverage-report.js, validated locally end-to-end at 38.7% (9605/24851 lines) on the existing record-traffic surface. Distribution sanity-checks: PostgresStorageAdapter 7.7% (mongo selected, postgres unused), MongoStorageAdapter 46%, RestQuery 52.8%, RestWrite 45.4%, Auth 40.5%, LiveQuery 3% (not exercised) — the numbers reflect what traffic actually executes.

Critical: the base Dockerfile and docker-compose.yml are untouched. Coverage instrumentation lives in a separate overlay applied only by the standalone GH Actions coverage workflow. The keploy/enterprise compat lane consumes the base compose unchanged.

How it works:

The overlay's Dockerfile.coverage installs c8 and drops a coverage-entrypoint.js shim that registers process.on('SIGTERM', () => process.exit(0)). Without that, express's app.listen pins the event loop and compose stop would signal-kill node (exit 143) before V8's NODE_V8_COVERAGE writer ran.
The overlay sets NODE_V8_COVERAGE=/coverage. Each Node process writes a coverage-<pid>-<ts>.json dump on clean exit.
coverage-report.js reads V8's nested block-coverage ranges, resolves per-line execution count against the deepest enclosing range, and emits a Covered N/M (XX.X%) summary plus a c8-shaped json-summary. We don't use c8 report because c8 filters node_modules/** even with --include overrides, and parse-server lives entirely under node_modules/parse-server.

flow.sh coverage against the base compose (no overlay) is a graceful no-op.

Coverage gate

.github/workflows/parse-server-mongo.yml triggers ONLY on changes under parse-server-mongo/**. Build (PR HEAD) + release (base ref) coverage compared; PR fails if drop exceeds 1.0pp. Threshold overridable via vars.PARSE_COVERAGE_THRESHOLD. The first-PR escape hatch detects a missing helper script on the base ref and treats baseline as 0%.

Run modes

Smoke check (without keploy): docker compose up -d && bash flow.sh bootstrap 240 && bash flow.sh record-traffic — exactly what the keploy enterprise lane wraps.
Real coverage (without keploy): apply the overlay, run bootstrap + record-traffic + compose stop -t 30 parse-server + flow.sh coverage.
Concurrent matrix cell: env-vars (PARSE_PROJECT, PARSE_HOST_PORT, PARSE_NETWORK_SUBNET, …) all ${VAR:-default}-substituted so multiple cells can run in parallel.

See README for full commands.

Consumers

keploy/enterprise .woodpecker/parse-server-linux.yml — three-cell record/replay matrix that delegates compose + bootstrap + traffic to this sample.

Test plan

docker compose up -d boots mongo + parse-server cleanly with no env vars
flow.sh bootstrap 240 signs up the fixed user, persists session token
flow.sh record-traffic exercises 51 fires across the surface above
Coverage validated locally end-to-end: 38.7% line coverage of node_modules/parse-server/lib
flow.sh coverage against base compose exits 0 cleanly with INFO message
Concurrent matrix cells confirmed via env-var overrides

…e template Make the parse-server-mongo sample directly consumable as a keploy compat-lane sample, on par with samples-typescript/umami-postgres and samples-python/ doccano-django. - flow.sh refactored to subcommand form: bootstrap | record-traffic | coverage | list-routes. bootstrap is idempotent (4xx during signup is treated as already-exists), persists the session token to /tmp/parse-token-${PARSE_PHASE}, and preserves the 3-second post-/health pause that the boot-phase _SCHEMA divergence reproducer needs. - record-traffic ports the broad parse-server REST + GraphQL surface (51 fire calls across users/sessions/classes/roles/files/cloud-functions/ schemas/hooks/graphql/aggregate). Every curl is preceded by log_fired so PARSE_FIRED_ROUTES_FILE captures (method,route) for coverage. The multi-class _SCHEMA mutation (GameScore, PlayerStats, Achievement) is retained so the boot-phase tiebreaker reproducer still triggers. - coverage subcommand reports (method,route) coverage with numerator from keploy/test-set-*/tests/*.yaml when present, else from the fired-routes log. Denominator from a curated 35-route table in parse_list_routes. - docker-compose.yml takes ${VAR:-default} for project name, network name, network subnet, IPs, container names, host:container port, and the internal PORT env. Defaults preserve standalone `docker compose up`; overrides let concurrent matrix cells share a single docker daemon. - keploy.yml.template carries the parse-server noise filter (header.Date, body.objectId, body.sessionToken, body.createdAt, body.updatedAt) so a lane consumer can apply it to keploy.yml after `keploy config --generate`. - README rewritten to document the four subcommands, all env vars, the route surface covered, the boot-phase divergence rationale, and a concurrent-cell run recipe. Signed-off-by: Akash Kumar <meakash7902@gmail.com>

Copilot

Pull request overview

Promotes parse-server-mongo/ from a minimal boot-phase reproducer into a keploy compat-lane sample by adding a subcommand-driven traffic/coverage driver, making compose configuration override-friendly for CI matrix concurrency, and providing a Keploy noise-filter template + updated docs.

Changes:

Reworked flow.sh into bootstrap | record-traffic | coverage | list-routes, broadened the fired API surface, and added a coverage report mode.
Parameterized docker-compose.yml using ${VAR:-default} for project/network/container naming, IPs, and ports to support concurrent runs on a shared daemon.
Added keploy.yml.template for Parse Server noise filtering and expanded README.md to document usage and overrides.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 7 comments.

File	Description
parse-server-mongo/flow.sh	Introduces subcommands, broader traffic generation, fired-route logging, and coverage reporting.
parse-server-mongo/docker-compose.yml	Makes compose names/network/IPs/ports/env configurable via environment variable defaults.
parse-server-mongo/keploy.yml.template	Adds a noise filter template for volatile Parse Server fields.
parse-server-mongo/README.md	Updates documentation for new flow/compose knobs and lane usage.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

+  case "${signup_status}" in
+    2*) echo "[flow] signup ok (HTTP ${signup_status})" ;;
+    4*)
+      # 4xx during bootstrap is treated as already-exists; re-run case.
+      echo "[flow] signup returned HTTP ${signup_status} — treating as already-exists"
+      ;;


+  token="$(printf '%s' "${login_resp}" | jq -r '.sessionToken // empty' 2>/dev/null || true)"
+  if [ -z "${token}" ]; then
+    echo "[flow] no sessionToken in login response — bootstrap will continue but session-bound calls will be skipped"
+    : > "${PARSE_TOKEN_FILE}"
+  else
+    printf '%s' "${token}" > "${PARSE_TOKEN_FILE}"


+# urlenc — minimal URL-encoder for query payloads.
+urlenc() {
+  jq -rn --arg v "$1" '$v|@uri'
+}


+      PARSE_SERVER_MASTER_KEY: ${PARSE_MASTER_KEY:-keploy-parse-master}
+      PARSE_SERVER_MASTER_KEY_IPS: ${PARSE_MASTER_KEY_IPS:-0.0.0.0/0,::0}
+      PARSE_SERVER_SERVER_URL: ${PARSE_SERVER_URL:-http://localhost:6100/parse}
+      PARSE_SERVER_DATABASE_URI: ${PARSE_DATABASE_URI:-mongodb://172.30.0.10:27017/parse}


 # parse-server-mongo

-Minimal Parse Server + MongoDB sample. Exists primarily as a falsifying e2e reproducer for the mongo/v2 boot-phase candidate-selection bug — a class of bug where an application issues the same query repeatedly during recording while DB state mutates, and the matcher's score-tied tiebreaker at replay picks the wrong same-shape mock.
+Parse Server (`parseplatform/parse-server` 9.x via `parse-server` npm 8.2.x) backed by MongoDB 7. Packaged as a complete keploy compat-lane sample: subcommand-driven traffic, env-driven docker-compose so concurrent matrix cells share a daemon, a noise-filter template that masks the parse-server identifiers that change every run, and a curated route list for coverage reporting.


+# fire METHOD URL [curl args...] — log then send a tolerant curl.
+fire() {
+  local method="$1"
+  local url="$2"
+  shift 2
+  log_fired "${method}" "${url}"
+  curl -sS -X "${method}" "${url}" "$@" >/dev/null 2>&1 || true


+      method="$(awk '/^    method:/{print $2; exit}' "${f}")"
+      route="$(awk '/^    url:/{print $2; exit}' "${f}")"


Introduce a GitHub Actions workflow scoped to the parse-server-mongo sample that runs the same end-to-end record/coverage shape as the doccano-django and umami-postgres lanes. The workflow is path-filtered to `parse-server-mongo/**` (plus the workflow file and its helper script) so it does NOT trigger on changes to other samples in this repo or on changes elsewhere. Three jobs: * build-coverage — runs the sample end-to-end against the PR's HEAD ref; emits a coverage percentage. * release-coverage — same, against the PR's base ref; first-PR bootstrap escape hatch returns 0% if the sample doesn't exist on the base ref yet. * coverage-gate — fails the PR if HEAD coverage drops more than COVERAGE_THRESHOLD percentage points below base. Default 1.0pp; override via the repo variable PARSE_COVERAGE_THRESHOLD. A sticky PR comment surfaces the base-vs-PR coverage diff with a prose explanation of which parse-server REST + GraphQL routes the sample's flow.sh::parse_record_traffic exercises. Helper script .github/workflows/scripts/run-and-measure-parse-server.sh is per-sample (sibling samples will land their own run-and-measure-<sample>.sh on parallel branches). It does docker compose up -d --build, polls /parse/health, runs flow.sh bootstrap → record-traffic → coverage with PARSE_FIRED_ROUTES_FILE as the standalone numerator, parses the report, and emits coverage=PCT to GITHUB_OUTPUT. This gate runs ONLY on PRs touching parse-server-mongo/. The enterprise PR pipeline (.woodpecker/parse-server-linux.yml) calls flow.sh coverage informationally and does NOT gate; the gate lives here on the sample repo, isolated from the enterprise lane. Signed-off-by: Akash Kumar <meakash7902@gmail.com>

…se ref parse-server-mongo/ already exists on main (PRs #94/#95), so the prior detect step returned true on every PR's release-coverage job and the measure step then tried to invoke a helper script that didn't exist on the base ref. Extending detect to require the helper script too lets the first-PR escape hatch fire when the workflow itself is the change being introduced. Signed-off-by: Akash Kumar <meakash7902@gmail.com>

github-actions · 2026-05-01T06:32:53Z

parse-server-mongo sample coverage

ref	coverage
base (`main`)	0.0%
this PR	38.7%

Threshold: PR may not drop coverage by more than 1.0pp. Override per-repo via the PARSE_COVERAGE_THRESHOLD actions variable.

Replaces the prior API-route-surface "coverage" (counting fired routes / curated route table) with actual V8 line coverage of the parse-server library code under node_modules/parse-server/lib. Architecture: - `Dockerfile.coverage` extends the base image with a graceful- shutdown shim (coverage-entrypoint.js installs SIGTERM/SIGINT handlers calling process.exit(0)) so V8 actually flushes its coverage data on `compose stop`. Without that shim, express's app.listen pins the loop and the kernel signal-kills node (exit 143) before NODE_V8_COVERAGE writes anything. - `docker-compose.coverage.yml` is an OVERLAY: applied via `-f docker-compose.yml -f docker-compose.coverage.yml`. It sets NODE_V8_COVERAGE=/coverage and bind-mounts the host ./coverage dir. The base `Dockerfile` and `docker-compose.yml` are untouched, so keploy/integrations and keploy/enterprise CI lanes consume the base compose and pay zero coverage cost. - `coverage-report.js` reads V8's per-process JSON dumps and produces a `Covered N/M (XX.X%)` summary plus a c8-shaped coverage-summary.json. We don't use `c8 report` because c8 filters node_modules even when --include is set, and parse-server lives entirely under node_modules/parse-server. The custom tool walks V8's nested ranges (block coverage) and resolves per-line execution count to the deepest range. - `flow.sh::parse_coverage` shells into the coverage image to run coverage-report.js against the dumped V8 data; when called against the base image (no overlay) it prints "INFO: ... uninstrumented" and exits 0 so enterprise lanes' `flow.sh coverage || true` informational calls keep working. Removed: - `parse_list_routes` (curated route table denominator). - `parse_collect_recorded_routes` (keploy-tests / fired-routes reader). - The legacy route-surface `parse_coverage` body. - `list-routes` subcommand. Validated locally: helper produced `coverage=38.7` to GITHUB_OUTPUT against a fresh stack (signup + record-traffic + clean stop). 38.7% reflects coverage of node_modules/parse-server/lib — mongo paths ~46%, REST ~52%, Auth ~40%, schema ~58%; postgres adapter ~7% (mongo storage selected) and LiveQuery ~3% (not exercised), which is the expected distribution. Signed-off-by: Akash Kumar <meakash7902@gmail.com>

Signed-off-by: Akash Kumar <meakash7902@gmail.com>

…/FIRED_ROUTES refs Signed-off-by: Akash Kumar <meakash7902@gmail.com>

Copilot AI review requested due to automatic review settings May 1, 2026 01:19

Copilot started reviewing on behalf of AkashKumar7902 May 1, 2026 01:20 View session

Copilot AI reviewed May 1, 2026

View reviewed changes

AkashKumar7902 added 2 commits May 1, 2026 11:45

officialasishkumar approved these changes May 1, 2026

View reviewed changes

AkashKumar7902 added 2 commits May 1, 2026 13:35

ci(parse-server-mongo): drop trailing prose from sticky comment

9926793

Signed-off-by: Akash Kumar <meakash7902@gmail.com>

docs(parse-server-mongo): document coverage overlay; drop list-routes…

c3df64a

…/FIRED_ROUTES refs Signed-off-by: Akash Kumar <meakash7902@gmail.com>

AkashKumar7902 changed the title ~~feat(parse-server-mongo): make sample compat-lane ready~~ feat(parse-server-mongo): compat-lane-ready sample + JS line coverage gate May 1, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(parse-server-mongo): compat-lane-ready sample + JS line coverage gate#97

feat(parse-server-mongo): compat-lane-ready sample + JS line coverage gate#97
AkashKumar7902 wants to merge 6 commits intomainfrom
feat/parse-server-mongo-compat-lane-ready

AkashKumar7902 commented May 1, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

github-actions Bot commented May 1, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		method="$(awk '/^ method:/{print $2; exit}' "${f}")"
		route="$(awk '/^ url:/{print $2; exit}' "${f}")"

Conversation

AkashKumar7902 commented May 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Layout

Coverage architecture

Coverage gate

Run modes

Consumers

Test plan

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

github-actions Bot commented May 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

parse-server-mongo sample coverage

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

AkashKumar7902 commented May 1, 2026 •

edited

Loading

github-actions Bot commented May 1, 2026 •

edited

Loading