feat: add sccache metrics to db

.github/actions/docker-build/action.yml

@@ -289,11 +289,13 @@ runs:
 
         # Run parser with all discovered stages
         # Usage: parse_buildkit_output.py <output_json> <stage1_name:log_file> [stage2_name:log_file] ... [--metadata=<file>]
+        # Note: sccache stats are parsed directly from the build logs (from 'use-sccache.sh show-stats' output)
         set +e
         python3 .github/scripts/parse_buildkit_output.py \
           "$COMPREHENSIVE_JSON" \
           "${STAGE_ARGS[@]}" \
-          "--metadata=${CONTAINER_METADATA}"
+          "--metadata=${CONTAINER_METADATA}" \
+          --debug
         PARSER_EXIT_CODE=$?
         set -e
 
@@ -315,4 +317,3 @@ runs:
         name: build-metrics-${{ inputs.framework }}-${{ inputs.target }}-${{ env.PLATFORM_ARCH }}-${{ github.run_id }}-${{ job.check_run_id }}
         path: build-metrics/build-${{ inputs.framework }}-${{ env.PLATFORM_ARCH }}-${{ github.run_id }}-${{ job.check_run_id }}.json
         retention-days: 7
-

.github/scripts/parse_buildkit_output.py

@@ -5,6 +5,8 @@
 Parse BuildKit output to extract detailed step-by-step metadata.
 BuildKit provides rich information about each build step including timing,
 cache status, sizes, and layer IDs.
+
+Also parses sccache statistics from the build log output.
 """
 
 import json
@@ -14,6 +16,113 @@
 from typing import Any, Dict, List
 
 
+def parse_sccache_from_log(log_content: str, debug: bool = False) -> Dict[str, Any]:
+    """
+    Parse sccache statistics from build log output.
+
+    In BuildKit logs, lines are prefixed with step numbers like:
+    #43 103.6 === sccache statistics AFTER Dynamo ===
+    #43 103.6 Compile requests                    2097
+    #43 103.6 Cache hits                          1670
+    #43 103.6 Cache hits rate                   100.00 %
+    """
+    sccache_data = {}
+
+    # Find sccache statistics section(s) - get the last one
+    # The section ends at the next DONE/CACHED marker or end of content
+    sections = re.findall(
+        r"=== sccache statistics AFTER ([^=]+) ===(.*?)(?=#\d+\s+DONE|#\d+\s+CACHED|#\d+\s+\[|$)",
+        log_content,
+        re.DOTALL,
+    )
+
+    if not sections:
+        if debug:
+            print("DEBUG: No sccache sections found in log", file=sys.stderr)
+        return {}
+
+    # Use the last sccache section (final stats)
+    build_name, stats_block = sections[-1]
+    sccache_data["build_name"] = build_name.strip()
+
+    if debug:
+        print(
+            f"DEBUG: Found sccache section for '{build_name.strip()}'", file=sys.stderr
+        )
+        print(
+            f"DEBUG: Stats block (first 1000 chars):\n{stats_block[:1000]}",
+            file=sys.stderr,
+        )
+
+    # Parse each statistic line
+    for line in stats_block.split("\n"):
+        line = line.strip()
+        if not line:
+            continue
+
+        # Remove BuildKit prefix if present: "#43 103.6 " or "#43 "
+        line = re.sub(r"^#\d+\s+[\d.]+\s+", "", line)
+        line = re.sub(r"^#\d+\s+", "", line)
+        line = line.strip()
+
+        if not line:
+            continue
+
+        # Skip section headers like "Non-cacheable reasons:"
+        if line.endswith(":"):
+            if debug:
+                print(f"DEBUG: Skipping section header: '{line}'", file=sys.stderr)
+            continue
+
+        # Match pattern: "Key Name    Value" or "Key Name    Value unit"
+        # Examples:
+        #   Compile requests                    2097
+        #   Cache hits (C/C++)                   890
+        #   Non-cacheable calls                  411
+        #   Cache hits rate                   100.00 %
+        #   Average cache read hit             0.050 s
+        #
+        # Note: there may be a space before the unit: "100.00 %" or "0.050 s"
+        # Key can contain: letters, numbers, (), /, space, hyphen, plus
+        match = re.match(
+            r"^([A-Za-z][A-Za-z0-9() /+\-]+?)\s{2,}([\d.]+)\s*(%|s)?\s*$", line
+        )
+        if match:
+            key_raw = match.group(1).strip()
+            value_str = match.group(2)
+            unit = match.group(3)
+
+            # Convert key to snake_case
+            key = key_raw.lower()
+            key = re.sub(r"[+()]", "", key)  # Remove plus and parentheses
+            key = re.sub(r"[/\-\s]+", "_", key)  # Replace /, -, spaces with _
+            key = re.sub(r"_+", "_", key)  # Collapse multiple underscores
+            key = key.strip("_")  # Remove leading/trailing underscores
+
+            # Add unit suffix
+            if unit == "%":
+                key = key + "_percent"
+            elif unit == "s":
+                key = key + "_seconds"
+
+            # Convert value to number
+            try:
+                if "." in value_str:
+                    sccache_data[key] = float(value_str)
+                else:
+                    sccache_data[key] = int(value_str)
+            except ValueError:
+                sccache_data[key] = value_str
+
+            if debug:
+                print(f"DEBUG: Parsed: {key} = {sccache_data[key]}", file=sys.stderr)
+        elif debug and line and any(c.isalpha() for c in line):
+            # Print non-matching lines in debug mode for troubleshooting
+            print(f"DEBUG: No match for line: '{line}'", file=sys.stderr)
+
+    return sccache_data
+
+
 class BuildKitParser:
     """Parser for BuildKit output logs"""
 
@@ -187,9 +296,12 @@ def main():
     # Parse arguments to find stage logs and metadata
     stage_logs = []  # List of (stage_name, log_file) tuples
     container_metadata_file = None
+    debug_mode = "--debug" in sys.argv
 
     for arg in sys.argv[2:]:
-        if arg.startswith("--metadata="):
+        if arg == "--debug":
+            continue
+        elif arg.startswith("--metadata="):
             container_metadata_file = arg.split("=", 1)[1]
         elif ":" in arg:
             stage_name, log_file = arg.split(":", 1)
@@ -209,6 +321,7 @@ def main():
     total_steps = 0
     total_cached = 0
     total_size = 0
+    sccache_data = {}
 
     # Parse each stage log
     for stage_name, log_file in stage_logs:
@@ -235,6 +348,15 @@ def main():
             total_cached += stage_data["container"]["cached_steps"]
             total_size += stage_data["container"]["total_size_transferred_bytes"]
 
+            # Extract sccache stats from the log (last one wins)
+            log_sccache = parse_sccache_from_log(log_content, debug=debug_mode)
+            if log_sccache:
+                sccache_data = log_sccache
+                print(
+                    f"✅ Found sccache stats in {stage_name} log: {len(log_sccache)} metrics",
+                    file=sys.stderr,
+                )
+
             print(
                 f"✅ Parsed {stage_name} stage: {stage_data['container']['total_steps']} steps",
                 file=sys.stderr,
@@ -274,6 +396,16 @@ def main():
         except Exception as e:
             print(f"Warning: Could not read container metadata: {e}", file=sys.stderr)
 
+    # Add sccache statistics (parsed from build logs)
+    if sccache_data:
+        build_data["container"]["sccache"] = sccache_data
+        print(
+            f"✅ sccache metrics added to container: {len(sccache_data)} metrics",
+            file=sys.stderr,
+        )
+    else:
+        print("ℹ️  No sccache stats found in build logs", file=sys.stderr)
+
     # Output JSON
     try:
         with open(output_json, "w") as f:
@@ -298,6 +430,25 @@ def main():
         file=sys.stderr,
     )
 
+    # Print sccache summary if available
+    if "sccache" in container:
+        sccache = container["sccache"]
+        print("", file=sys.stderr)
+        print("🔨 sccache Summary:", file=sys.stderr)
+        if "compile_requests" in sccache:
+            print(
+                f"   Compile Requests: {sccache['compile_requests']}", file=sys.stderr
+            )
+        if "cache_hits" in sccache:
+            print(f"   Cache Hits: {sccache['cache_hits']}", file=sys.stderr)
+        if "cache_misses" in sccache:
+            print(f"   Cache Misses: {sccache['cache_misses']}", file=sys.stderr)
+        if "cache_hits_rate_percent" in sccache:
+            print(
+                f"   Cache Hit Rate: {sccache['cache_hits_rate_percent']:.2f}%",
+                file=sys.stderr,
+            )
+
 
 if __name__ == "__main__":
     main()

container/use-sccache.sh

@@ -15,9 +15,10 @@ usage() {
 Usage: $0 [COMMAND] [OPTIONS]
 
 Commands:
-    install         Install sccache binary (requires ARCH_ALT environment variable)
-    show-stats      Display sccache statistics with optional build name
-    help            Show this help message
+    install             Install sccache binary (requires ARCH_ALT environment variable)
+    show-stats          Display sccache statistics with optional build name
+    export-stats-json   Export sccache statistics as JSON to a file
+    help                Show this help message
 
 Environment variables:
     USE_SCCACHE             Set to 'true' to enable sccache
@@ -31,6 +32,8 @@ Examples:
     ARCH_ALT=x86_64 $0 install
     # Show stats with build name
     $0 show-stats "UCX"
+    # Export stats to JSON file
+    $0 export-stats-json /path/to/output.json "BuildName"
 EOF
 }
 
@@ -59,6 +62,95 @@ show_stats() {
     fi
 }
 
+export_stats_json() {
+    local output_file="${1:-/tmp/sccache-stats.json}"
+    local build_name="${2:-unknown}"
+
+    if ! command -v sccache >/dev/null 2>&1; then
+        # Create empty stats file so COPY commands don't fail
+        echo '{"build_name": "'"$build_name"'", "sccache_available": false}' > "$output_file"
+        echo "ℹ️  sccache not available, created placeholder: $output_file"
+        return 0  # Don't fail the build
+    fi
+
+    # Get raw stats output
+    local stats_output
+    stats_output=$(sccache --show-stats 2>&1)
+
+    # Parse the output and convert to JSON using awk
+    echo "$stats_output" | awk -v build_name="$build_name" '
+    BEGIN {
+        print "{"
+        print "  \"build_name\": \"" build_name "\","
+        first = 1
+    }
+
+    # Match lines with format: "Key    Value" or "Key    Value %"
+    /^[A-Za-z]/ {
+        # Extract key and value
+        key = ""
+        value = ""
+        unit = ""
+
+        # Find the position where multiple spaces start (separator between key and value)
+        for (i = 1; i <= NF; i++) {
+            if ($i ~ /^[0-9.]+$/) {
+                # Found the start of the value
+                value = $i
+                # Check if there'\''s a unit after the value
+                if (i + 1 <= NF && ($( i+1) == "%" || $(i+1) == "s")) {
+                    unit = $(i+1)
+                }
+                # Everything before this is the key
+                for (j = 1; j < i; j++) {
+                    if (key == "") {
+                        key = $j
+                    } else {
+                        key = key " " $j
+                    }
+                }
+                break
+            }
+        }
+
+        if (key != "" && value != "") {
+            # Convert key to snake_case
+            gsub(/[()]/, "", key)  # Remove parentheses
+            gsub(/\//, "_", key)   # Replace / with _
+            gsub(/ /, "_", key)    # Replace spaces with _
+            key = tolower(key)
+
+            # Add unit suffix if applicable
+            if (unit == "%") {
+                key = key "_percent"
+            } else if (unit == "s") {
+                key = key "_seconds"
+            }
+
+            # Check if value is a valid JSON number (integer or single decimal)
+            # Version strings like "0.8.2" need to be quoted
+            is_number = (value ~ /^[0-9]+$/ || value ~ /^[0-9]+\.[0-9]+$/)
+
+            # Print JSON field
+            if (!first) print ","
+            if (is_number) {
+                printf "  \"%s\": %s", key, value
+            } else {
+                printf "  \"%s\": \"%s\"", key, value
+            }
+            first = 0
+        }
+    }
+
+    END {
+        print ""
+        print "}"
+    }
+    ' > "$output_file"
+
+    echo "✅ sccache stats exported to: $output_file"
+}
+
 main() {
     case "${1:-help}" in
         install)
@@ -72,6 +164,10 @@ main() {
             shift  # Remove the command from arguments
             show_stats "$@"  # Pass all remaining arguments
             ;;
+        export-stats-json)
+            shift  # Remove the command from arguments
+            export_stats_json "$@"  # Pass all remaining arguments
+            ;;
         help|--help|-h)
             usage
             ;;