Class: Vector::Function::FlexibleDownScaling

Inherits:
Object
  • Object
show all
Includes:
HLogger
Defined in:
lib/vector/functions/flexible_down_scaling.rb

Instance Method Summary collapse

Methods included from HLogger

enable, #hlog, #hlog_ctx, #hlog_ctx_string

Constructor Details

#initialize(options) ⇒ FlexibleDownScaling

Returns a new instance of FlexibleDownScaling.



8
9
10
11
12
13
14
15
16
17
18
19
20
21
 
# File 'lib/vector/functions/flexible_down_scaling.rb', line 8

def initialize(options)
  @cloudwatch = options[:cloudwatch]
  @dry_run = options[:dry_run]
  @up_down_cooldown = options[:up_down_cooldown]
  @down_down_cooldown = options[:down_down_cooldown]
  @max_sunk_cost = options[:max_sunk_cost]
  @variable_thresholds = options[:variable_thresholds]
  @n_low = options[:n_low]
  @n_high = options[:n_high]
  @m = options[:m]
  @g_low = options[:g_low]
  @g_high = options[:g_high]
  @debug_variable_thresholds = options[:print_variable_thresholds]
end

Instance Method Details

#run_for(group, ps_check_procs) ⇒ Object 



23
24
25
26
27
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
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
 
# File 'lib/vector/functions/flexible_down_scaling.rb', line 23

def run_for(group, ps_check_procs)
  result = { :triggered => false }

  hlog_ctx("fds") do
    hlog_ctx("group:#{group.name}") do
      # don't check if no config was specified
      if @up_down_cooldown.nil? && @down_down_cooldown.nil?
        hlog("No cooldown periods specified, exiting")
        return result
      end

      # don't bother checking for a scaledown if desired capacity is
      # already at the minimum size...
      if group.desired_capacity == group.min_size
        hlog("Group is already at minimum size, exiting")
        return result
      end

      scaledown_policies = group.scaling_policies.select do |policy|
        policy.scaling_adjustment < 0
      end

      scaledown_policies.each do |policy|
        hlog_ctx("policy:#{policy.name}") do
          # TODO: support adjustment types other than ChangeInCapacity here
          if policy.adjustment_type == "ChangeInCapacity" &&
             ps_check_procs &&
             ps_check_procs.any? {|ps_check_proc|
               ps_check_proc.call(group.desired_capacity + policy.scaling_adjustment, self) }
            hlog("Predictive scaleup would trigger a scaleup if group were shrunk")
            next
          end

          alarms = policy.alarms.keys.map do |alarm_name|
            @cloudwatch.alarms[alarm_name]
          end

          # only consider disabled alarms (enabled alarms will trigger
          # the policy automatically)
          disabled_alarms = alarms.select do |alarm|
            !alarm.enabled?
          end

          # Do this logic first in case the user is just trying to print out
          # the thresholds.
          if @variable_thresholds
            # variable_thresholds currently requires a CPUUtilization alarm to function
            vt_cpu_alarm = disabled_alarms.find {|alarm| alarm.metric_name == "CPUUtilization" }

            # remove this alarm from the check, since we're not checking its alarm status
            # below.
            disabled_alarms.delete(vt_cpu_alarm)

            unless vt_cpu_alarm
              hlog("Variable thresholds requires an alarm on CPUUtilization, skipping")
              next
            end

            @n_low ||= group.min_size + 1
            @n_high ||= group.max_size
            @m ||= vt_cpu_alarm.threshold / 100

            if @g_low == @g_high
              hlog("g_low == g_high (#{@g_low}), not attempting to use flexible thresholds.")
              next
            end

            if @n_low == @n_high
              hlog("n_low == n_high (#{@n_low}), not attempting to use flexible thresholds.")
              next
            end

            if @debug_variable_thresholds
              puts "  n_low: #{@n_low}"
              puts " n_high: #{@n_high}"
              puts "      m: #{@m}"
              puts "  g_low: #{@g_low}"
              puts " g_high: #{@g_high}"
              puts
              puts "  N  Threshold"
              ([@n_low, group.min_size].min + 1).upto([@n_high, group.max_size].max) do |i|
                puts " %2d  %.1f%%" % [i, (variable_threshold(i, @n_low, @n_high, @m, @g_low, @g_high) * 100)]
              end
              next
            end
          end

          unless disabled_alarms.all? {|alarm| alarm.state_value == "ALARM" }
            hlog("Not all alarms are in ALARM state")
            next
          end

          if @variable_thresholds
            threshold = variable_threshold(group.desired_capacity, @n_low, @n_high, @m, @g_low, @g_high)

            stats = vt_cpu_alarm.metric.statistics(
              :start_time => Time.now - (vt_cpu_alarm.period * vt_cpu_alarm.evaluation_periods),
              :end_time => Time.now,
              :statistics => [ vt_cpu_alarm.statistic ],
              :period => vt_cpu_alarm.period)

            if stats.datapoints.length < vt_cpu_alarm.evaluation_periods
              hlog("Could not get enough datapoints for checking variable threshold");
              next
            end

            if stats.datapoints.any? {|dp| dp[vt_cpu_alarm.statistic.downcase.to_sym] > (threshold * 100) }
              hlog("Not all datapoints are beneath the variable threshold #{(threshold * 100).to_i}: #{stats.datapoints}")
              next
            end

            hlog("Variable threshold: #{(threshold * 100).to_i}, #{group.desired_capacity} nodes")
          end

          unless outside_cooldown_period(group)
            hlog("Group is not outside the specified cooldown periods")
            next
          end

          unless has_eligible_scaledown_instance(group)
            hlog("Group does not have an instance eligible for scaledown due to max_sunk_cost")
            next
          end

          if @dry_run
            hlog("Executing policy (DRY RUN)")
          else
            hlog("Executing policy")
            policy.execute(:honor_cooldown => true)
          end

          result[:triggered] = true

          # no need to evaluate other scaledown policies
          return result
        end
      end
    end
  end

  result
end