Class: RailsErrorDashboard::Services::CascadeDetector

Inherits:
Object
  • Object
show all
Defined in:
lib/rails_error_dashboard/services/cascade_detector.rb

Overview

Detects cascade patterns by analyzing error occurrences

Runs periodically to find errors that consistently follow other errors, indicating a causal relationship.

Constant Summary collapse

DETECTION_WINDOW =

Time window to look for cascades (errors within this window may be related)

60.seconds
MIN_CASCADE_FREQUENCY =

Minimum times a pattern must occur to be considered a cascade

3
MIN_CASCADE_PROBABILITY =

Minimum probability threshold (% of time parent leads to child)

0.7

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(lookback_hours: 24) ⇒ CascadeDetector

Returns a new instance of CascadeDetector.



23
24
25
26
 
# File 'lib/rails_error_dashboard/services/cascade_detector.rb', line 23

def initialize(lookback_hours: 24)
  @lookback_hours = lookback_hours
  @detected_count = 0
end

Class Method Details

.call(lookback_hours: 24) ⇒ Object 



19
20
21
 
# File 'lib/rails_error_dashboard/services/cascade_detector.rb', line 19

def self.call(lookback_hours: 24)
  new(lookback_hours: lookback_hours).detect_cascades
end

Instance Method Details

#detect_cascades ⇒ Object 



28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
 
# File 'lib/rails_error_dashboard/services/cascade_detector.rb', line 28

def detect_cascades
  return { detected: 0, updated: 0 } unless can_detect?

  # Pluck (error_log_id, occurred_at) for every occurrence in the window
  # ordered chronologically. Using pluck instead of loading full
  # ActiveRecord rows keeps memory bounded to ~16 bytes/row instead of
  # ~5KB/row, which matters because the host app schedules this job and
  # the lookback window may contain a lot of occurrences.
  start_time = @lookback_hours.hours.ago
  rows = ErrorOccurrence
    .where("occurred_at >= ?", start_time)
    .order(:occurred_at)
    .pluck(:error_log_id, :occurred_at)

  # Two-pointer sweep: occurrences are time-sorted, so for each parent we
  # only advance the child pointer forward through occurrences within the
  # detection window. O(N + pairs) instead of O(N) inner SQL queries.
  patterns_found = Hash.new { |h, k| h[k] = { delays: [], count: 0 } }

  rows.each_with_index do |(parent_id, parent_time), i|
    window_end = parent_time + DETECTION_WINDOW
    j = i + 1
    while j < rows.length
      child_id, child_time = rows[j]
      break if child_time > window_end

      # Match the original SQL `occurred_at > parent` — strict, so two
      # occurrences with identical timestamps don't form a cascade pair.
      if child_id != parent_id && child_time > parent_time
        key = [ parent_id, child_id ]
        patterns_found[key][:delays] << (child_time - parent_time).to_f
        patterns_found[key][:count] += 1
      end
      j += 1
    end
  end

  # Filter and persist cascade patterns via Command
  updated_count = 0
  patterns_found.each do |(parent_id, child_id), data|
    next if data[:count] < MIN_CASCADE_FREQUENCY

    avg_delay = data[:delays].sum / data[:delays].size

    result = Commands::UpsertCascadePattern.call(
      parent_error_id: parent_id,
      child_error_id: child_id,
      frequency: data[:count],
      avg_delay_seconds: avg_delay
    )

    if result[:created]
      @detected_count += 1
    else
      updated_count += 1
    end
  end

  { detected: @detected_count, updated: updated_count }
end