FINERACT-2181: Use matrix jobs via Github Actions to improve performance

scripts/split-features.sh

@@ -0,0 +1,110 @@
+#!/bin/bash
+
+# Exit on any error
+set -e
+
+# Get the number of shards (default to 10 if not provided)
+NUM_SHARDS=${1:-10}
+# Convert to 0-based index for internal calculations
+SHARD_INDEX_ZERO_BASED=$((${2:-1} - 1))
+SHARD_INDEX=${2:-1}  # Keep original 1-based index for output
+
+# Directory containing feature files
+FEATURES_DIR="fineract-e2e-tests-runner/src/test/resources/features"
+TEMP_FILE="/tmp/feature_scenarios_$(date +%s).txt"
+
+# Check if features directory exists
+if [ ! -d "$FEATURES_DIR" ]; then
+  echo "Error: Features directory not found at $FEATURES_DIR"
+  exit 1
+fi
+
+# Function to count scenarios in a feature file
+count_scenarios() {
+  local file="$1"
+  # Count scenario and scenario outline keywords, excluding commented lines
+  grep -v '^[[:space:]]*#' "$file" | grep -c 'Scenario\( Outline\)\?:' || echo "0"
+}
+
+# Process each feature file and count scenarios
+echo "Analyzing feature files to count scenarios..."
+> "$TEMP_FILE"
+
+while IFS= read -r -d $'\0' file; do
+  # Remove the 'fineract-e2e-tests-runner/' prefix
+  rel_path="${file#fineract-e2e-tests-runner/}"
+  scenario_count=$(count_scenarios "$file")
+  echo "$scenario_count $rel_path"
+done < <(find "$FEATURES_DIR" -type f -name '*.feature' -print0) | sort -nr > "$TEMP_FILE"
+
+# Read the sorted list of features
+SORTED_FEATURES=()
+while IFS= read -r line; do
+  # Extract just the file path (removing the scenario count)
+  path=$(echo "$line" | awk '{$1=""; print $0}' | sed 's/^ //')
+  [ -n "$path" ] && SORTED_FEATURES+=("$path")
+done < "$TEMP_FILE"
+TOTAL_FEATURES=${#SORTED_FEATURES[@]}
+
+# Check if any feature files were found
+if [ $TOTAL_FEATURES -eq 0 ]; then
+  echo "Warning: No feature files found in $FEATURES_DIR"
+  # Create an empty feature list file
+  FEATURE_LIST_FILE="feature_shard_${SHARD_INDEX}.txt"
+  > "$FEATURE_LIST_FILE"
+  echo "Created empty feature list file: $FEATURE_LIST_FILE"
+  rm -f "$TEMP_FILE"
+  exit 0
+fi
+
+# Create a file to store the feature file paths
+# Use the 1-based index for the filename
+FEATURE_LIST_FILE="feature_shard_${SHARD_INDEX}.txt"
+> "$FEATURE_LIST_FILE"
+
+# Distribute features to shards in a round-robin fashion to balance scenario counts
+for ((i=0; i<TOTAL_FEATURES; i++)); do
+  if (( i % NUM_SHARDS == SHARD_INDEX_ZERO_BASED )); then
+    echo "${SORTED_FEATURES[$i]}" >> "$FEATURE_LIST_FILE"
+  fi
+done
+
+# Count scenarios in this shard
+SHARD_SCENARIOS=0
+while IFS= read -r line; do
+  [ -z "$line" ] && continue
+  count=$(grep -m 1 -F "$line" "$TEMP_FILE" | awk '{print $1}')
+  SHARD_SCENARIOS=$((SHARD_SCENARIOS + count))
+done < "$FEATURE_LIST_FILE"
+
+# Get the list of features in this shard for output
+SHARD_FEATURES=()
+while IFS= read -r line; do
+  [ -n "$line" ] && SHARD_FEATURES+=("$line")
+done < "$FEATURE_LIST_FILE"
+NUM_FEATURES=${#SHARD_FEATURES[@]}
+
+# Output the shard information
+echo "Shard $SHARD_INDEX (1-based): $NUM_FEATURES features with $SHARD_SCENARIOS total scenarios"
+if [ $NUM_FEATURES -gt 0 ]; then
+  echo "First feature: ${SHARD_FEATURES[0]}"
+  if [ $NUM_FEATURES -gt 1 ]; then
+    echo "Last feature: ${SHARD_FEATURES[$((NUM_FEATURES-1))]}"
+  fi
+  echo "Features in this shard (scenario count):"
+  while IFS= read -r line; do
+    [ -z "$line" ] && continue
+    count=$(grep -m 1 -F "$line" "$TEMP_FILE" | awk '{print $1}')
+    echo "  - $line ($count scenarios)"
+  done < "$FEATURE_LIST_FILE"
+fi
+
+echo "Feature list written to $FEATURE_LIST_FILE"
+
+# Clean up temp file
+rm -f "$TEMP_FILE"
+
+# Set output for GitHub Actions
+if [ -n "$GITHUB_OUTPUT" ]; then
+  echo "FEATURE_LIST_FILE=$FEATURE_LIST_FILE" >> $GITHUB_OUTPUT
+fi

scripts/split-tests.sh

@@ -0,0 +1,52 @@
+#!/bin/bash
+# Usage: ./split-tests.sh <total-shards> <shard-index>
+set -e
+
+TOTAL_SHARDS=$1
+SHARD_INDEX=$2
+
+if [[ -z "$TOTAL_SHARDS" || -z "$SHARD_INDEX" ]]; then
+  echo "ERROR: You must provide <total-shards> and <shard-index>."
+  exit 1
+fi
+
+echo "🔍 Searching for eligible JUnit test classes..."
+
+ALL_TESTS=$(find . -type f -path "*/src/test/java/*Test.java" \
+  | while read filepath; do
+      filename=$(basename "$filepath")
+
+      # Skip abstract class or interface by name
+      if [[ "$filename" =~ ^Abstract.*Test\.java$ || "$filename" =~ .*AbstractTest\.java$ ]]; then
+        echo "Skipping abstract-named file: $filename" >&2
+        continue
+      fi
+
+      # Skip if file doesn't contain valid JUnit annotations
+      if ! grep -qE "@Test|@Nested|@ParameterizedTest" "$filepath"; then
+        continue
+      fi
+
+      # Extract module directory path (everything before /src/test/java)
+      module_path="${filepath%%/src/test/java/*}"
+      # Convert from ./custom/acme/loan/job to :custom:acme:loan:job
+      module_name=$(echo "$module_path" | sed 's|^\./||; s|/|:|g; s|^|:|')
+
+      # Extract fully qualified test class name
+      class_name=$(echo "$filepath" | sed 's|^.*src/test/java/||; s|/|.|g; s|.java$||')
+
+      echo "$module_name,$class_name"
+    done \
+  | sort)
+
+TOTAL_COUNT=$(echo "$ALL_TESTS" | wc -l)
+echo "Found $TOTAL_COUNT eligible test classes."
+
+SELECTED_CLASSES=$(echo "$ALL_TESTS" \
+  | awk -v ts="$TOTAL_SHARDS" -v si="$SHARD_INDEX" 'NR % ts == (si - 1)')
+
+OUTPUT_FILE="shard-tests_${SHARD_INDEX}.txt"
+echo "$SELECTED_CLASSES" > "$OUTPUT_FILE"
+
+echo "Selected $(wc -l < "$OUTPUT_FILE") classes for shard $SHARD_INDEX of $TOTAL_SHARDS:"
+cat "$OUTPUT_FILE"