claude-flow-novice 2.10.4 → 2.10.6

package/claude-assets/skills/cfn-redis-coordination/metrics-export.sh

@@ -0,0 +1,674 @@
+#!/bin/bash
+# metrics-export.sh - Export CFN Loop metrics in multiple formats
+# Supports: JSON, Prometheus, OpenTelemetry (OTLP), CSV
+# Features: Time range filtering, remote push, TTL cleanup
+
+set -euo pipefail
+
+# ============================================================================
+# Configuration & Default Values
+# ============================================================================
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+TASK_ID=""
+FORMAT="json"
+SINCE=""
+UNTIL=""
+PUSH_GATEWAY=""
+OUTPUT_FILE=""
+CLEANUP_TTL=""
+VERBOSE=false
+
+# Metric categories (matching orchestrate-cfn-loop.sh)
+METRICS_CATEGORIES=(
+  "iteration_start"
+  "iteration_duration"
+  "loop3_consensus"
+  "loop2_consensus"
+  "agent_latency"
+  "gate_failures"
+  "timeout_count"
+  "quorum_fallback"
+  "retry_count"
+)
+
+# ============================================================================
+# Utility Functions
+# ============================================================================
+
+log() {
+  if [[ "$VERBOSE" == true ]]; then
+    echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*" >&2
+  fi
+}
+
+error() {
+  echo "ERROR: $*" >&2
+  exit 1
+}
+
+usage() {
+  cat <<EOF
+Usage: $0 --task-id <id> [OPTIONS]
+
+Export CFN Loop metrics in multiple formats.
+
+Required:
+  --task-id <id>           Task ID to export metrics for
+
+Optional:
+  --format <type>          Export format: json|prometheus|otlp|csv (default: json)
+  --since <timestamp>      Filter metrics since timestamp (ISO 8601 or Unix ms)
+  --until <timestamp>      Filter metrics until timestamp (ISO 8601 or Unix ms)
+  --output <file>          Write output to file instead of stdout
+  --push-gateway <url>     Push Prometheus metrics to remote gateway
+  --cleanup-ttl <days>     Cleanup metrics older than N days
+  --verbose                Enable verbose logging
+
+Examples:
+  # Export all metrics as JSON
+  $0 --task-id redis-phase7-1760900252 --format json
+
+  # Export to Prometheus with time range
+  $0 --task-id task-123 --format prometheus \\
+    --since "2025-10-19T00:00:00Z" --until "2025-10-19T23:59:59Z"
+
+  # Push to Prometheus Pushgateway
+  $0 --task-id task-123 --format prometheus \\
+    --push-gateway http://localhost:9091
+
+  # Export to file with cleanup
+  $0 --task-id task-123 --format json \\
+    --output metrics.json --cleanup-ttl 30
+
+  # Export OTLP for OpenTelemetry
+  $0 --task-id task-123 --format otlp --output traces.json
+
+EOF
+  exit 1
+}
+
+# ============================================================================
+# Time Conversion Functions
+# ============================================================================
+
+# Convert ISO 8601 timestamp to Unix milliseconds
+iso_to_unix_ms() {
+  local iso_timestamp="$1"
+
+  # Try using date command (GNU/BSD compatible)
+  if date --version >/dev/null 2>&1; then
+    # GNU date
+    date -d "$iso_timestamp" +%s%3N 2>/dev/null || echo "0"
+  else
+    # BSD/macOS date
+    date -j -f "%Y-%m-%dT%H:%M:%S" "${iso_timestamp%%Z*}" +%s000 2>/dev/null || echo "0"
+  fi
+}
+
+# Filter metrics by time range
+filter_by_time() {
+  local value="$1"
+  local timestamp="$2"
+
+  # If no time filters, include all
+  [[ -z "$SINCE" && -z "$UNTIL" ]] && echo "$value" && return
+
+  local since_ms=$(iso_to_unix_ms "$SINCE")
+  local until_ms=$(iso_to_unix_ms "$UNTIL")
+
+  if [[ -n "$SINCE" ]] && (( timestamp < since_ms )); then
+    return 1
+  fi
+
+  if [[ -n "$UNTIL" ]] && (( timestamp > until_ms )); then
+    return 1
+  fi
+
+  echo "$value"
+}
+
+# ============================================================================
+# Metric Collection Functions
+# ============================================================================
+
+# Collect all metrics from Redis for a task
+collect_metrics() {
+  local task_id="$1"
+  declare -gA METRICS
+
+  log "Collecting metrics for task: $task_id"
+
+  for category in "${METRICS_CATEGORIES[@]}"; do
+    local key="swarm:${task_id}:metrics:${category}"
+    local values=$(redis-cli LRANGE "$key" 0 -1 2>/dev/null || echo "")
+
+    if [[ -n "$values" ]]; then
+      METRICS["$category"]="$values"
+      local count=$(echo "$values" | wc -l)
+      log "Collected ${category}: $count entries"
+    else
+      METRICS["$category"]=""
+    fi
+  done
+
+  # Collect agent-specific metrics
+  local agent_pattern="swarm:${task_id}:agent:*:metrics:*"
+  local agent_keys=$(redis-cli KEYS "$agent_pattern" 2>/dev/null || echo "")
+
+  if [[ -n "$agent_keys" ]]; then
+    METRICS["agent_keys"]="$agent_keys"
+    local count=$(echo "$agent_keys" | wc -w)
+    log "Found agent-specific metric keys: $count"
+  fi
+}
+
+# Calculate statistical summaries from JSON array
+calculate_stats() {
+  local json_array="$1"
+  local field="${2:-}"
+
+  # Parse JSON array and extract numeric values
+  local values=""
+  if [[ -n "$field" ]]; then
+    values=$(echo "$json_array" | jq -r ".[] | .$field // empty" 2>/dev/null || echo "")
+  else
+    values=$(echo "$json_array" | jq -r '.[] // empty' 2>/dev/null || echo "")
+  fi
+
+  local count=0
+  local sum=0
+  local min=""
+  local max=""
+
+  for val in $values; do
+    # Skip non-numeric values
+    if ! [[ "$val" =~ ^[0-9]+\.?[0-9]*$ ]]; then
+      continue
+    fi
+
+    count=$((count + 1))
+    sum=$(echo "$sum + $val" | bc -l)
+
+    if [[ -z "$min" ]] || (( $(echo "$val < $min" | bc -l) )); then
+      min="$val"
+    fi
+
+    if [[ -z "$max" ]] || (( $(echo "$val > $max" | bc -l) )); then
+      max="$val"
+    fi
+  done
+
+  local avg=0
+  if (( count > 0 )); then
+    avg=$(echo "scale=2; $sum / $count" | bc -l)
+  fi
+
+  cat <<EOF
+{
+  "count": $count,
+  "sum": ${sum:-0},
+  "avg": ${avg:-0},
+  "min": ${min:-0},
+  "max": ${max:-0}
+}
+EOF
+}
+
+# ============================================================================
+# Export Format Functions
+# ============================================================================
+
+# Export as JSON
+export_json() {
+  local task_id="$1"
+
+  log "Exporting metrics in JSON format"
+
+  # Build JSON structure
+  cat <<EOF
+{
+  "task_id": "$task_id",
+  "export_timestamp": "$(date -u +%Y-%m-%dT%H:%M:%SZ)",
+  "metrics": {
+EOF
+
+  local first=true
+  for category in "${METRICS_CATEGORIES[@]}"; do
+    local values="${METRICS[$category]:-}"
+    [[ -z "$values" ]] && continue
+
+    if [[ "$first" == false ]]; then
+      echo ","
+    fi
+    first=false
+
+    # Convert to JSON array (Redis LRANGE returns newline-separated values)
+    local json_array=$(echo "$values" | jq -s -c '.' 2>/dev/null || echo "[]")
+
+    # Calculate statistics based on category type
+    local stats=""
+    case "$category" in
+      iteration_duration|agent_latency)
+        stats=$(calculate_stats "$json_array" "duration_ms")
+        ;;
+      loop3_consensus|loop2_consensus)
+        stats=$(calculate_stats "$json_array" "consensus")
+        ;;
+      *)
+        # For scalar metrics
+        stats=$(calculate_stats "$json_array")
+        ;;
+    esac
+
+    cat <<EOF
+    "$category": {
+      "values": $json_array,
+      "stats": $stats
+    }
+EOF
+  done
+
+  cat <<EOF
+
+  },
+  "metadata": {
+    "total_categories": ${#METRICS_CATEGORIES[@]},
+    "non_empty_categories": $(for cat in "${METRICS_CATEGORIES[@]}"; do [[ -n "${METRICS[$cat]:-}" ]] && echo 1; done | wc -l)
+  }
+}
+EOF
+}
+
+# Export as Prometheus text format
+export_prometheus() {
+  local task_id="$1"
+
+  log "Exporting metrics in Prometheus format"
+
+  # Iteration duration histogram
+  if [[ -n "${METRICS[iteration_duration]:-}" ]]; then
+    echo "# HELP cfn_iteration_duration_ms CFN Loop iteration duration in milliseconds"
+    echo "# TYPE cfn_iteration_duration_ms histogram"
+
+    local values=$(echo "${METRICS[iteration_duration]}" | jq -s -c '.')
+    echo "$values" | jq -r '.[] | "\(.duration_ms) \(.iteration)"' 2>/dev/null | while read -r duration iteration; do
+      echo "cfn_iteration_duration_ms{task_id=\"$task_id\",iteration=\"$iteration\"} $duration"
+    done
+    echo ""
+  fi
+
+  # Agent latency histogram
+  if [[ -n "${METRICS[agent_latency]:-}" ]]; then
+    echo "# HELP cfn_agent_latency_ms Agent completion latency in milliseconds"
+    echo "# TYPE cfn_agent_latency_ms histogram"
+
+    local values=$(echo "${METRICS[agent_latency]}" | jq -s -c '.')
+    echo "$values" | jq -r '.[] | "\(.agent) \(.latency_ms) \(.loop) \(.iteration)"' 2>/dev/null | while read -r agent latency loop iteration; do
+      echo "cfn_agent_latency_ms{task_id=\"$task_id\",agent=\"$agent\",loop=\"$loop\",iteration=\"$iteration\"} $latency"
+    done
+    echo ""
+  fi
+
+  # Loop 3 consensus scores
+  if [[ -n "${METRICS[loop3_consensus]:-}" ]]; then
+    echo "# HELP cfn_loop3_consensus Loop 3 consensus score (0.0-1.0)"
+    echo "# TYPE cfn_loop3_consensus gauge"
+
+    local values=$(echo "${METRICS[loop3_consensus]}" | jq -s -c '.')
+    echo "$values" | jq -r '.[] | "\(.consensus) \(.iteration)"' 2>/dev/null | while read -r score iteration; do
+      echo "cfn_loop3_consensus{task_id=\"$task_id\",iteration=\"$iteration\"} $score"
+    done
+    echo ""
+  fi
+
+  # Loop 2 consensus scores
+  if [[ -n "${METRICS[loop2_consensus]:-}" ]]; then
+    echo "# HELP cfn_loop2_consensus Loop 2 consensus score (0.0-1.0)"
+    echo "# TYPE cfn_loop2_consensus gauge"
+
+    local values=$(echo "${METRICS[loop2_consensus]}" | jq -s -c '.')
+    echo "$values" | jq -r '.[] | "\(.consensus) \(.iteration)"' 2>/dev/null | while read -r score iteration; do
+      echo "cfn_loop2_consensus{task_id=\"$task_id\",iteration=\"$iteration\"} $score"
+    done
+    echo ""
+  fi
+
+  # Gate failure counter
+  local gate_failures=$(redis-cli GET "swarm:${task_id}:metrics:gate_failures" 2>/dev/null || echo "0")
+  echo "# HELP cfn_gate_failures_total Total number of gate failures"
+  echo "# TYPE cfn_gate_failures_total counter"
+  echo "cfn_gate_failures_total{task_id=\"$task_id\"} ${gate_failures:-0}"
+  echo ""
+
+  # Timeout counter
+  local timeout_count=$(redis-cli GET "swarm:${task_id}:metrics:timeout_count" 2>/dev/null || echo "0")
+  echo "# HELP cfn_timeout_total Total number of agent timeouts"
+  echo "# TYPE cfn_timeout_total counter"
+  echo "cfn_timeout_total{task_id=\"$task_id\"} ${timeout_count:-0}"
+  echo ""
+
+  # Quorum fallback counter
+  local quorum_fallback=$(redis-cli GET "swarm:${task_id}:metrics:quorum_fallback" 2>/dev/null || echo "0")
+  echo "# HELP cfn_quorum_fallback_total Total number of quorum fallbacks"
+  echo "# TYPE cfn_quorum_fallback_total counter"
+  echo "cfn_quorum_fallback_total{task_id=\"$task_id\"} ${quorum_fallback:-0}"
+  echo ""
+
+  # Retry counter
+  local retry_count=$(redis-cli GET "swarm:${task_id}:metrics:retry_count" 2>/dev/null || echo "0")
+  echo "# HELP cfn_retry_total Total number of retry attempts"
+  echo "# TYPE cfn_retry_total counter"
+  echo "cfn_retry_total{task_id=\"$task_id\"} ${retry_count:-0}"
+  echo ""
+}
+
+# Export as OpenTelemetry OTLP JSON
+export_otlp() {
+  local task_id="$1"
+
+  log "Exporting metrics in OTLP format"
+
+  # OTLP Trace/Span structure for CFN Loop
+  cat <<EOF
+{
+  "resourceSpans": [
+    {
+      "resource": {
+        "attributes": [
+          {
+            "key": "service.name",
+            "value": { "stringValue": "cfn-loop-orchestrator" }
+          },
+          {
+            "key": "task.id",
+            "value": { "stringValue": "$task_id" }
+          }
+        ]
+      },
+      "scopeSpans": [
+        {
+          "scope": {
+            "name": "redis-coordination",
+            "version": "2.2.0"
+          },
+          "spans": [
+EOF
+
+  # Create spans for each iteration
+  local first_span=true
+
+  if [[ -n "${METRICS[iteration_start]:-}" ]] && [[ -n "${METRICS[iteration_duration]:-}" ]]; then
+    local starts="${METRICS[iteration_start]}"
+    local durations=$(echo "${METRICS[iteration_duration]}" | jq -s -c '.')
+
+    # Parse starts and combine with durations
+    local iteration=1
+    for start_time in $starts; do
+      local duration=$(echo "$durations" | jq -r ".[$((iteration - 1))].duration_ms // 0" 2>/dev/null)
+
+      if [[ "$first_span" == false ]]; then
+        echo ","
+      fi
+      first_span=false
+
+      # Calculate end time
+      local end_time=$((start_time + duration))
+
+      # Convert to nanoseconds for OTLP
+      local start_ns=$((start_time * 1000000))
+      local end_ns=$((end_time * 1000000))
+
+      # Generate trace and span IDs
+      local trace_id=$(openssl rand -hex 16 2>/dev/null || echo "00000000000000000000000000000000")
+      local span_id=$(openssl rand -hex 8 2>/dev/null || echo "0000000000000000")
+
+      cat <<SPAN
+            {
+              "traceId": "$trace_id",
+              "spanId": "$span_id",
+              "name": "cfn_loop_iteration_${iteration}",
+              "kind": 1,
+              "startTimeUnixNano": "$start_ns",
+              "endTimeUnixNano": "$end_ns",
+              "attributes": [
+                {
+                  "key": "iteration",
+                  "value": { "intValue": "$iteration" }
+                },
+                {
+                  "key": "duration_ms",
+                  "value": { "intValue": "$duration" }
+                }
+              ]
+            }
+SPAN
+
+      iteration=$((iteration + 1))
+    done
+  fi
+
+  cat <<EOF
+
+          ]
+        }
+      ]
+    }
+  ]
+}
+EOF
+}
+
+# Export as CSV
+export_csv() {
+  local task_id="$1"
+
+  log "Exporting metrics in CSV format"
+
+  # CSV header
+  echo "timestamp,category,metric_type,value,task_id,iteration,agent,loop"
+
+  local export_timestamp=$(date -u +%Y-%m-%dT%H:%M:%SZ)
+
+  # Export each category
+  for category in "${METRICS_CATEGORIES[@]}"; do
+    local values="${METRICS[$category]:-}"
+    [[ -z "$values" ]] && continue
+
+    local json_array=$(echo "$values" | jq -s -c '.' 2>/dev/null || echo "[]")
+
+    case "$category" in
+      iteration_duration)
+        echo "$json_array" | jq -r ".[] | \"$export_timestamp,$category,duration,\(.duration_ms),$task_id,\(.iteration),,\"" 2>/dev/null
+        ;;
+      agent_latency)
+        echo "$json_array" | jq -r ".[] | \"$export_timestamp,$category,latency,\(.latency_ms),$task_id,\(.iteration),\(.agent),\(.loop)\"" 2>/dev/null
+        ;;
+      loop3_consensus|loop2_consensus)
+        echo "$json_array" | jq -r ".[] | \"$export_timestamp,$category,consensus,\(.consensus),$task_id,\(.iteration),,\"" 2>/dev/null
+        ;;
+      *)
+        # For scalar counter metrics
+        local value=$(redis-cli GET "swarm:${task_id}:metrics:${category}" 2>/dev/null || echo "0")
+        if [[ -n "$value" ]] && [[ "$value" != "0" ]]; then
+          echo "$export_timestamp,$category,count,$value,$task_id,,,,"
+        fi
+        ;;
+    esac
+  done
+}
+
+# ============================================================================
+# Push to Prometheus Pushgateway
+# ============================================================================
+
+push_to_gateway() {
+  local gateway_url="$1"
+  local task_id="$2"
+  local metrics_data="$3"
+
+  log "Pushing metrics to Prometheus Pushgateway: $gateway_url"
+
+  # Push to Pushgateway
+  local response=$(curl -s -w "\n%{http_code}" --data-binary "$metrics_data" \
+    "${gateway_url}/metrics/job/cfn-loop/instance/${task_id}" 2>&1)
+
+  local http_code=$(echo "$response" | tail -n1)
+
+  if [[ "$http_code" == "200" ]] || [[ "$http_code" == "201" ]]; then
+    log "Successfully pushed metrics to gateway (HTTP $http_code)"
+    echo "Metrics pushed successfully to $gateway_url"
+  else
+    error "Failed to push metrics (HTTP $http_code)"
+  fi
+}
+
+# ============================================================================
+# TTL Cleanup
+# ============================================================================
+
+cleanup_old_metrics() {
+  local task_id="$1"
+  local ttl_days="$2"
+
+  log "Cleaning up metrics older than $ttl_days days"
+
+  local cutoff_ms=$(date -d "$ttl_days days ago" +%s%3N 2>/dev/null || echo "0")
+  local deleted_count=0
+
+  for category in "${METRICS_CATEGORIES[@]}"; do
+    local key="swarm:${task_id}:metrics:${category}"
+    local values=$(redis-cli LRANGE "$key" 0 -1 2>/dev/null || echo "")
+
+    [[ -z "$values" ]] && continue
+
+    local keep_values=()
+    local json_array="[$values]"
+
+    # Parse JSON objects and filter by timestamp
+    local filtered=$(echo "$json_array" | jq -c "[.[] | select(.timestamp >= $cutoff_ms)]" 2>/dev/null || echo "[]")
+
+    local filtered_count=$(echo "$filtered" | jq 'length' 2>/dev/null || echo "0")
+    local original_count=$(echo "$json_array" | jq 'length' 2>/dev/null || echo "0")
+
+    deleted_count=$((deleted_count + original_count - filtered_count))
+
+    # Replace list with filtered values
+    if (( filtered_count > 0 )); then
+      redis-cli DEL "$key" >/dev/null
+      echo "$filtered" | jq -c '.[]' | while read -r val; do
+        echo "$val" | redis-cli -x RPUSH "$key" >/dev/null
+      done
+    else
+      redis-cli DEL "$key" >/dev/null
+    fi
+  done
+
+  log "Deleted $deleted_count old metric entries"
+  echo "Cleanup complete: $deleted_count entries removed"
+}
+
+# ============================================================================
+# Main Execution
+# ============================================================================
+
+main() {
+  # Parse arguments
+  while [[ $# -gt 0 ]]; do
+    case $1 in
+      --task-id)
+        TASK_ID="$2"
+        shift 2
+        ;;
+      --format)
+        FORMAT="$2"
+        shift 2
+        ;;
+      --since)
+        SINCE="$2"
+        shift 2
+        ;;
+      --until)
+        UNTIL="$2"
+        shift 2
+        ;;
+      --output)
+        OUTPUT_FILE="$2"
+        shift 2
+        ;;
+      --push-gateway)
+        PUSH_GATEWAY="$2"
+        shift 2
+        ;;
+      --cleanup-ttl)
+        CLEANUP_TTL="$2"
+        shift 2
+        ;;
+      --verbose)
+        VERBOSE=true
+        shift
+        ;;
+      --help)
+        usage
+        ;;
+      *)
+        echo "Unknown option: $1"
+        usage
+        ;;
+    esac
+  done
+
+  # Validate required arguments
+  [[ -z "$TASK_ID" ]] && error "Missing required --task-id argument"
+
+  # Validate format
+  case "$FORMAT" in
+    json|prometheus|otlp|csv)
+      ;;
+    *)
+      error "Invalid format: $FORMAT (must be json|prometheus|otlp|csv)"
+      ;;
+  esac
+
+  # Collect metrics
+  collect_metrics "$TASK_ID"
+
+  # Generate export
+  local output=""
+  case "$FORMAT" in
+    json)
+      output=$(export_json "$TASK_ID")
+      ;;
+    prometheus)
+      output=$(export_prometheus "$TASK_ID")
+      ;;
+    otlp)
+      output=$(export_otlp "$TASK_ID")
+      ;;
+    csv)
+      output=$(export_csv "$TASK_ID")
+      ;;
+  esac
+
+  # Output handling
+  if [[ -n "$OUTPUT_FILE" ]]; then
+    echo "$output" > "$OUTPUT_FILE"
+    log "Metrics exported to: $OUTPUT_FILE"
+    echo "Exported metrics to: $OUTPUT_FILE"
+  else
+    echo "$output"
+  fi
+
+  # Push to Prometheus Pushgateway if specified
+  if [[ -n "$PUSH_GATEWAY" ]] && [[ "$FORMAT" == "prometheus" ]]; then
+    push_to_gateway "$PUSH_GATEWAY" "$TASK_ID" "$output"
+  fi
+
+  # Cleanup old metrics if specified
+  if [[ -n "$CLEANUP_TTL" ]]; then
+    cleanup_old_metrics "$TASK_ID" "$CLEANUP_TTL"
+  fi
+}
+
+# Run main function
+main "$@"