fix: update stock variance account logic which defaults to default expense #40365

Workflow file for this run

.github/workflows/server-tests-mariadb.yml at 5a1c43b

	name: Server (Mariadb)

	on:
	repository_dispatch:
	types: [frappe-framework-change]
	pull_request:
	paths-ignore:
	- '**.js'
	- '**.css'
	- '**.svg'
	- '**.md'
	- '**.html'
	- 'crowdin.yml'
	- '.coderabbit.yml'
	- '.mergify.yml'
	schedule:
	# Run everday at midnight UTC / 5:30 IST
	- cron: "0 0 * * *"
	workflow_dispatch:
	inputs:
	user:
	description: 'Frappe Framework repository user (add your username for forks)'
	required: true
	default: 'frappe'
	type: string
	branch:
	description: 'Frappe Framework branch'
	default: 'develop'
	required: false
	type: string

	permissions:
	contents: read
	packages: read

	concurrency:
	group: server-mariadb-develop-${{ github.event_name }}-${{ github.event.number \|\| github.event_name == 'workflow_dispatch' && github.run_id \|\| '' }}
	cancel-in-progress: true

	# Shared across both jobs. Both run in the SAME CI image so the bench lives at the identical
	# path (/home/ci/frappe-bench) on the setup runner and the test shards — that's what makes the
	# packaged Python venv portable between them.
	env:
	TZ: 'Asia/Kolkata'
	DEBIAN_FRONTEND: noninteractive
	NODE_ENV: "production"
	WITH_COVERAGE: ${{ github.event_name != 'pull_request' }}
	ERPNEXT_CI_USER: ci
	PIP_CACHE_DIR: /home/ci/.cache/pip
	npm_config_cache: /home/ci/.cache/npm
	YARN_CACHE_FOLDER: /home/ci/.cache/yarn
	UV_CACHE_DIR: /home/ci/.cache/uv

	jobs:
	# Build the bench (clone + pip + yarn + assets) and reinstall test_site ONCE, on a free
	# GitHub-hosted runner, then publish the whole bench (with a DB dump baked in) as an artifact.
	# The expensive, non-parallelisable work happens here exactly once instead of on every shard.
	setup:
	name: Build & reinstall (setup)
	# Dedicated scale set (fat cpu request) so the build+reinstall runs at full speed, uncontended
	# by the many thin test shards. Same CI image + /home/ci path + 127.0.0.1 DB as the shards,
	# so the packaged bench (and its venv) transplants cleanly.
	runs-on: erpnext-arc-setup
	timeout-minutes: 40
	container:
	image: ghcr.io/frappe/erpnext-ci-mariadb:py3.14-node24
	credentials:
	username: ${{ secrets.GHCR_USERNAME \|\| github.actor }}
	password: ${{ secrets.GHCR_TOKEN \|\| github.token }}
	defaults:
	run:
	shell: bash
	steps:
	- name: Clone
	uses: actions/checkout@v6

	- name: Check for valid Python & Merge Conflicts
	run: \|
	python -m compileall -fq "${GITHUB_WORKSPACE}"
	if grep -lr --exclude-dir=node_modules "^<<<<<<< " "${GITHUB_WORKSPACE}"
	then echo "Found merge conflicts"
	exit 1
	fi

	- name: Add to Hosts
	run: echo "127.0.0.1 test_site" \| sudo tee -a /etc/hosts

	# MariaDB runs in-container on a datadir OUTSIDE the bench, because install.sh's next step
	# does `rm -rf ~/frappe-bench`. After the reinstall, the datadir is moved into the bench so
	# it ships in the artifact — test shards then start an already-loaded server (no restore).
	- name: Start DB
	run: bash ${GITHUB_WORKSPACE}/.github/helper/start-db.sh
	env:
	SKIP_SYSTEM_SETUP: "1"
	CI_DB_DATADIR: /home/ci/db-data

	- name: Install
	run: bash ${GITHUB_WORKSPACE}/.github/helper/install.sh
	env:
	DB: mariadb
	TYPE: server
	FRAPPE_USER: ${{ github.event.inputs.user }}
	FRAPPE_BRANCH: ${{ github.event.client_payload.sha \|\| github.event.inputs.branch }}
	DB_HOST: 127.0.0.1
	DB_USER_HOST: '%'
	WKHTMLTOX_DEB: /tmp/wkhtmltox.deb
	SKIP_SYSTEM_SETUP: "1"
	SKIP_WKHTMLTOX_SETUP: "1"

	- name: Warm up test data
	run: \|
	su -m "${ERPNEXT_CI_USER:-frappe}" -s /bin/bash <<'EOF'
	cd ~/frappe-bench/
	bench --site test_site run-tests --lightmode --module erpnext.tests.bootstrap_test_data
	EOF

	# Clean shutdown (consistent InnoDB datadir), then stage it inside the bench for packaging.
	- name: Stop DB and stage datadir
	run: \|
	mariadb-admin -h 127.0.0.1 -P 3306 -u root -proot shutdown \|\| true
	for _ in $(seq 1 30); do [ -f /home/ci/db-data/mysqld.pid ] \|\| break; sleep 1; done
	# Don't bake a dirty datadir — fail if mariadbd didn't finish stopping, rather than ship
	# an inconsistent datadir the shards would have to crash-recover.
	[ -f /home/ci/db-data/mysqld.pid ] && { echo "mariadbd did not shut down cleanly"; exit 1; }
	mv /home/ci/db-data /home/ci/frappe-bench/mariadb-data

	# Package the whole bench (apps, venv, node_modules, sites, the DB dump, and hydrate.sh)
	# into one artifact for the test shards to consume.
	# Single-node hand-off: stage the bench on a node-local hostPath instead of round-tripping
	# through GitHub artifact storage (~60s/shard). Setup and shards share the same disk, so
	# the shards just untar it locally. NOTE: this assumes one node — a shard on a different
	# node could not read this path (then you'd need GitHub artifacts or an NFS/RWX volume).
	- name: Stage bench on node (hostPath)
	run: \|
	cp "${GITHUB_WORKSPACE}/.github/helper/hydrate.sh" /home/ci/frappe-bench/hydrate.sh
	cp "${GITHUB_WORKSPACE}/.github/helper/start-db.sh" /home/ci/frappe-bench/start-db.sh
	mkdir -p /opt/ci-bench-staging
	# self-clean: drop bench tars from runs older than 2h
	find /opt/ci-bench-staging -maxdepth 1 -name '*.tar.gz' -mmin +120 -delete 2>/dev/null \|\| true
	# Exclude .git/node_modules; the mariadb-data datadir IS included (the pre-loaded DB).
	tar czpf "/opt/ci-bench-staging/${GITHUB_RUN_ID}.tar.gz" -C /home/ci \
	--exclude='.git' --exclude='node_modules' frappe-bench
	ls -lh "/opt/ci-bench-staging/${GITHUB_RUN_ID}.tar.gz"

	# Fan-out: each shard downloads the bench, untars it, starts MariaDB on the baked datadir, and
	# runs its slice of the suite. No clone, no build, no reinstall, no DB dump restore on the shards.
	test:
	name: Python Unit Tests
	needs: setup
	runs-on: erpnext-arc
	timeout-minutes: 60
	container:
	image: ghcr.io/frappe/erpnext-ci-mariadb:py3.14-node24
	credentials:
	username: ${{ secrets.GHCR_USERNAME \|\| github.actor }}
	password: ${{ secrets.GHCR_TOKEN \|\| github.token }}
	defaults:
	run:
	shell: bash

	strategy:
	fail-fast: false
	matrix:
	container: [1, 2, 3, 4]

	steps:
	- name: Add to Hosts
	run: echo "127.0.0.1 test_site" \| sudo tee -a /etc/hosts

	# Read the bench straight from the node-local hostPath the setup job staged it on — no
	# GitHub download. -p preserves the ci (uid 1001) ownership so bench runs as ci cleanly.
	- name: Untar bench from node (hostPath)
	run: \|
	tar xzpf "/opt/ci-bench-staging/${GITHUB_RUN_ID}.tar.gz" -C /home/ci
	ls -ld /home/ci/frappe-bench

	- name: Hydrate (start DB on baked datadir + bench start)
	run: bash /home/ci/frappe-bench/hydrate.sh
	env:
	DB_HOST: 127.0.0.1
	SKIP_SYSTEM_SETUP: "1"

	- name: Run Tests
	run: \|
	su -m "${ERPNEXT_CI_USER:-frappe}" -s /bin/bash <<'EOF'
	cd ~/frappe-bench/
	coverage_flag=""
	if [ "$WITH_COVERAGE" = "true" ]; then coverage_flag="--with-coverage"; fi
	bench --site test_site run-parallel-tests --lightmode --app erpnext \
	--total-builds ${{ strategy.job-total }} \
	--build-number ${{ matrix.container }} \
	$coverage_flag
	EOF
	env:
	TYPE: server

	- name: Show bench output
	if: ${{ always() }}
	run: cat ~/frappe-bench/bench_start.log \|\| true

	- name: Upload coverage data
	if: ${{ env.WITH_COVERAGE == 'true' }}
	uses: actions/upload-artifact@v4
	with:
	name: coverage-${{ matrix.container }}
	path: /home/ci/frappe-bench/sites/coverage.xml

	coverage:
	name: Coverage Wrap Up
	needs: [test]
	if: ${{ github.event_name != 'pull_request' }}
	runs-on: ubuntu-latest
	steps:
	- name: Clone
	uses: actions/checkout@v6

	- name: Download artifacts
	uses: actions/download-artifact@v4

	- name: Upload coverage data
	uses: codecov/codecov-action@v4
	with:
	name: MariaDB
	token: ${{ secrets.CODECOV_TOKEN }}
	fail_ci_if_error: true
	verbose: true

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix: update stock variance account logic which defaults to default expense #40365

Workflow file

fix: update stock variance account logic which defaults to default expense #40365

Uh oh!

Workflow file for this run