Class: Atatus::Collector::Base

Inherits:
Object
  • Object
show all
Includes:
Logging
Defined in:
lib/atatus/collector/base.rb

Constant Summary

Constants included from Logging

Logging::LEVELS, Logging::PREFIX

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods included from Logging

#debug, #error, #fatal, #info, #warn

Constructor Details

#initialize(config) ⇒ Base

Returns a new instance of Base.



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
 
# File 'lib/atatus/collector/base.rb', line 27

def initialize(config)
  # info 'Initializing Collector'
  @config = config
  @spans = Hash.new {|h,k| h[k]=[]}

  @txns_lock = Mutex.new
  @txns_agg = {}
  @txn_hist_agg = {}
  @traces_agg = []
  @error_metrics_agg = {}
  @error_requests_agg = []

  @errors_lock = Mutex.new
  @errors_aggs = []

  @metrics_lock = Mutex.new
  @metrics_agg = []

  @analytics = @config.analytics
  @analytics_lock = Mutex.new
  @analytics_agg = []

  @hostinfo_response = {}
  @hostinfo_response["analytics"] = true
  @transport = Atatus::BaseTransport.new(config)
  @collect_counter = 0
  @running = false
end

Instance Attribute Details

#config ⇒ Object (readonly)

Returns the value of attribute config.



56
57
58
 
# File 'lib/atatus/collector/base.rb', line 56

def config
  @config
end

Instance Method Details

#add_analytics(txn) ⇒ Object 



162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
 
# File 'lib/atatus/collector/base.rb', line 162

def add_analytics(txn)
  analytics_txn = {}
  analytics_txn[:timestamp] = Util.ms(txn.timestamp).to_i
  analytics_txn[:txnId] = ""
  analytics_txn[:txnId] = txn.id if !txn.id.nil?
  analytics_txn[:traceId] = ""
  analytics_txn[:traceId] = txn.trace_id if !txn.trace_id.nil?
  analytics_txn[:name] = txn.name
  analytics_txn[:duration] = Util.ms(txn.duration)

  if !txn.context.nil?
    if !txn.context.request.nil?
      r = txn.context.request

      analytics_txn[:method] = r.method if !r.method.nil?

      if !r.headers.nil?
        analytics_txn[:requestHeaders] = r.headers
        analytics_txn[:userAgent] = r.headers['User-Agent'] if r.headers.key?('User-Agent')
      end

      if !r.body.nil?
        if !r.body.empty?
          if r.body != "[SKIPPED]"
            analytics_txn[:requestBody] = r.body.to_json
          end
        end
      end

      if !r.url.nil?
        analytics_txn[:url] = r.url.full
      end

      if !r.socket.nil?
        analytics_txn[:ip] = r.socket.remote_addr
      end

    end

    if
      defined?(txn.context.response_body) &&
      !txn.context.response_body.nil?
    then
      analytics_txn[:responseBody] = txn.context.response_body
    end

    if !txn.context.response.nil?
      if !txn.context.response.headers.nil?
        analytics_txn[:responseHeaders] = txn.context.response.headers
      end
      if !txn.context.response.status_code.nil?
        analytics_txn[:statusCode] = txn.context.response.status_code.to_i
      end
    end

    if
      defined?(txn.context.custom) &&
      !txn.context.custom.nil? &&
      !txn.context.custom.empty?
    then
      analytics_txn[:customData] = txn.context.custom
    end

    if
      defined?(txn.context.user) &&
      !txn.context.user.nil? &&
      !txn.context.user.empty?
    then
      analytics_txn[:userId] = txn.context.user.id
      analytics_txn[:userEmail] = txn.context.user.email
      analytics_txn[:userName] = txn.context.user.username
    end

    if
      defined?(txn.context.company) &&
      !txn.context.company.nil? &&
      !txn.context.company.empty?
    then
      analytics_txn[:companyId] = txn.context.company.id
    end
  end

  @analytics_lock.synchronize do
    if @analytics_agg.length < 10000
      @analytics_agg << analytics_txn
    end
  end
end

#add_error(error) ⇒ Object 



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
 
# File 'lib/atatus/collector/base.rb', line 86

def add_error(error)
  ensure_worker_running
  ignore_error = false
  if
    !error.exception.nil?  &&
    !error.exception.type.nil? &&
    !error.exception.message.nil?
  then
    if @hostinfo_response.key?("ignoreExceptionPatterns")
      @hostinfo_response['ignoreExceptionPatterns'].each do |k, v|
        if error.exception.type.match(k)
          exception_values = v
          if exception_values.length == 0
            ignore_error = true
          else
            exception_values.each do |value|
              if error.exception.message.include?(value)
                ignore_error = true
              end
            end
          end
          break
        end
      end
    end
  end
  if ignore_error == false
    @errors_lock.synchronize do
      if @errors_aggs.length < 20
        @errors_aggs.push(error)
      else
        i = rand(20)
        @errors_aggs[i] = error
      end
    end
  end
end

#add_metrics(metricset) ⇒ Object 



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
 
# File 'lib/atatus/collector/base.rb', line 124

def add_metrics(metricset)
  ensure_worker_running

  metric_added = false
  metric = {}

  if %i[system.cpu.total.norm.pct system.memory.actual.free system.memory.total system.process.cpu.total.norm.pct system.process.memory.size system.process.memory.rss.bytes].all? {|s| metricset.samples.key? s}
  then
    metric[:'system.cpu.total.norm.pct'] = metricset.samples[:'system.cpu.total.norm.pct']
    metric[:'system.memory.actual.free'] = metricset.samples[:'system.memory.actual.free']
    metric[:'system.memory.total'] = metricset.samples[:'system.memory.total']
    metric[:'system.process.cpu.total.norm.pct'] = metricset.samples[:'system.process.cpu.total.norm.pct']
    metric[:'system.process.memory.size'] = metricset.samples[:'system.process.memory.size']
    metric[:'system.process.memory.rss.bytes'] = metricset.samples[:'system.process.memory.rss.bytes']
    metric_added = true
  end

  if %i[ruby.gc.count ruby.threads ruby.heap.slots.live ruby.heap.slots.free ruby.heap.allocations.total].all? {|s| metricset.samples.key? s}
  then
    metric[:'ruby.gc.count'] = metricset.samples[:'ruby.gc.count']
    metric[:'ruby.threads'] = metricset.samples[:'ruby.threads']
    metric[:'ruby.heap.slots.live'] = metricset.samples[:'ruby.heap.slots.live']
    metric[:'ruby.heap.slots.free'] = metricset.samples[:'ruby.heap.slots.free']
    metric[:'ruby.heap.allocations.total'] = metricset.samples[:'ruby.heap.allocations.total']
    if %i[ruby.gc.time].all? {|s| metricset.samples.key? s}
    then
      metric[:'ruby.gc.time'] = metricset.samples[:'ruby.gc.time']
    end
    metric_added = true
  end

  if metric_added
    @metrics_lock.synchronize do
      @metrics_agg << metric
    end
  end
end

#add_span(span) ⇒ Object 



251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
 
# File 'lib/atatus/collector/base.rb', line 251

def add_span(span)
  ensure_worker_running

  if
    span.transaction_id.nil? ||
    span.name.nil? ||
    span.type.nil? ||
    span.subtype.nil? ||
    span.duration.nil?
  then
    return
  end

  @spans[span.transaction_id] << span if span.transaction_id
end

#add_txn(txn) ⇒ Object 



267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
 
# File 'lib/atatus/collector/base.rb', line 267

def add_txn(txn)
  ensure_worker_running

  if
    txn.name.nil? ||
    txn.id.nil? ||
    txn.timestamp.nil? ||
    txn.duration.nil?
  then
    return
  end

  ignore_txn = false
  if @hostinfo_response.key?("ignoreTxnNamePatterns")
    @hostinfo_response['ignoreTxnNamePatterns'].each do |k|
      if txn.name.match(k)
        ignore_txn = true
        break
      end
    end
  end
  if ignore_txn == true
    return
  end

  return if txn.name.empty?
  return if txn.duration <= 0

  @txns_lock.synchronize do
    txn_type = @config.framework_name || "Ruby"
    background = false
    if !txn.type.nil?
      if txn.type == "Sidekiq"
        background = true
      end
    end

    if !@txns_agg.key?(txn.name)
      @txns_agg[txn.name] = Txn.new(txn_type, "Ruby", txn.duration, background: background)
      @txns_agg[txn.name].id = txn.id
      @txns_agg[txn.name].pid = txn.id
    else
      @txns_agg[txn.name].aggregate! txn.duration
    end

    if background == false and txn.duration <= 150*1000*1000.0
      if !@txn_hist_agg.key?(txn.name)
        @txn_hist_agg[txn.name] = TxnHist.new(txn_type, "Ruby", Util.ms(txn.duration))
      else
        @txn_hist_agg[txn.name].aggregate! Util.ms(txn.duration)
      end
    end

    spans_present = false
    ruby_time = 0
    spans_tuple = []
    if @spans.key?(txn.id)
      spans_present = true
      @spans[txn.id].each do |span|
        if
          span.name.nil? ||
          span.type.nil? ||
          span.subtype.nil? ||
          span.timestamp.nil? ||
          span.duration.nil?
        then
          next
        end

        next if span.name.empty?

        if span.timestamp >= txn.timestamp
          start = Util.ms(span.timestamp - txn.timestamp)
          spans_tuple.push(SpanTiming.new(span.name, span.type, span.subtype, start, start + Util.ms(span.duration), span.duration, span.id, span.transaction_id))
        end
      end
    end

    if spans_tuple.length == 0
      ruby_time = Util.ms(txn.duration)
    else
      spans_tuple.sort! {| a, b | a[:start] <=> b[:start] }
      j = 0
      while j < spans_tuple.length
        if spans_tuple[j].subtype == 'controller' || spans_tuple[j].subtype == 'view' || spans_tuple[j].subtype == 'tilt'
          k = j+1
          while k < spans_tuple.length
            if spans_tuple[k].start >= spans_tuple[j].end
              break
            else
              if spans_tuple[k].end <= spans_tuple[j].end
                spans_tuple[j].duration -= Util.us(spans_tuple[k].end - spans_tuple[k].start)
              else
                spans_tuple[j].duration -= Util.us(spans_tuple[j].end - spans_tuple[k].start)
              end
            end
            k += 1
          end
          if spans_tuple[j].duration <= 0
            spans_tuple[j].duration = 1
          end
        end

        if !@txns_agg[txn.name].spans.key?(spans_tuple[j].name)
          kind = Layer.span_kind(spans_tuple[j].type)
          type = Layer.span_type(spans_tuple[j].subtype)
          @txns_agg[txn.name].spans[spans_tuple[j].name] = Layer.new(type, kind, spans_tuple[j].duration)
          @txns_agg[txn.name].spans[spans_tuple[j].name].id = spans_tuple[j].id
          @txns_agg[txn.name].spans[spans_tuple[j].name].pid = spans_tuple[j].transaction_id
        else
          @txns_agg[txn.name].spans[spans_tuple[j].name].aggregate! spans_tuple[j].duration
        end

        j += 1
      end

      ruby_time = spans_tuple[0].start
      span_end = spans_tuple[0].end
      j = 0
      while j < spans_tuple.length
        if spans_tuple[j].start > span_end
          ruby_time += spans_tuple[j].start - span_end
          span_end = spans_tuple[j].end
        else
          if spans_tuple[j].end > span_end
            span_end = spans_tuple[j].end
          end
        end
        j += 1
      end
      if Util.ms(txn.duration) > span_end
        ruby_time += Util.ms(txn.duration) - span_end
      end
    end

    if ruby_time > 0
      ruby_time = Util.us(ruby_time)
      @txns_agg[txn.name].spans["Ruby"] = Layer.new("Ruby", "Ruby", ruby_time)
    end

    if spans_present == true || ruby_time > 0
      if Util.ms(txn.duration) >= @config.trace_threshold
        trace_txn = txn
        trace_txn.spans = @spans[txn.id] if spans_present
        trace_txn.ruby_time = ruby_time if ruby_time > 0

        if @traces_agg.length < 5
          @traces_agg.push(trace_txn)
        else
          i = rand(5)
          @traces_agg[i] = trace_txn
        end
      end
    end

    if spans_present
      @spans.delete(txn.id)
    end

    if
      !txn.context.nil? &&
      !txn.context.response.nil? &&
      !txn.context.response.status_code.nil?
    then
      status_code = txn.context.response.status_code.to_i
      ignore_status_code = false
      if status_code >= 400 && status_code != 404
        if @hostinfo_response.key?("ignoreHTTPFailurePatterns")
          @hostinfo_response['ignoreHTTPFailurePatterns'].each do |k, v|
            if txn.name.match(k)
              status_code_array_s = v
              if status_code_array_s.length == 0
                ignore_status_code = true
              else
                status_code_array_s.each do |code|
                  if code == txn.context.response.status_code.to_s
                    ignore_status_code = true
                  end
                end
              end
              break
            end
          end
        end
        if ignore_status_code == false
          if !@error_metrics_agg.key?(txn.name)
            @error_metrics_agg[txn.name] = {status_code => 1}
          else
            if !@error_metrics_agg[txn.name].key?(status_code)
              @error_metrics_agg[txn.name][status_code] = 1
            else
              @error_metrics_agg[txn.name][status_code] += 1
            end
          end

          trace_id = ""
          trace_id = txn.trace_id if !txn.trace_id.nil?
          if @error_requests_agg.length < 20
            @error_requests_agg.push({'name' => txn.name, 'txn_id' => txn.id, 'trace_id' => trace_id, 'context' => txn.context})
          else
            i = rand(20)
            @error_requests_agg[i] = {'name' => txn.name, 'txn_id' => txn.id, 'trace_id' => trace_id, 'context' => txn.context}
          end
        end
      end
    end

    if @hostinfo_response.key?("analytics")
      if @hostinfo_response["analytics"] == true && @analytics == true
        if background == false
          add_analytics txn
        end
      end
    end
  end
end

#handle_forking! ⇒ Object 



81
82
83
84
 
# File 'lib/atatus/collector/base.rb', line 81

def handle_forking!
  stop
  start
end

#pid_str ⇒ Object 



58
59
60
 
# File 'lib/atatus/collector/base.rb', line 58

def pid_str
  format('[PID:%s]', Process.pid)
end

#start ⇒ Object 



62
63
64
65
66
 
# File 'lib/atatus/collector/base.rb', line 62

def start
  debug '%s: Starting collector', pid_str

  ensure_worker_running
end

#stop ⇒ Object 



68
69
70
71
72
73
74
75
76
77
78
79
 
# File 'lib/atatus/collector/base.rb', line 68

def stop
  return unless @running
  @running = false
  if worker_active?
    debug '%s: Waiting for collector worker to exit', pid_str
    @worker.run
    @worker.join(10)
  end
rescue => e
  error format('Failed during collector stop: [%s] %s', e.class, e.message)
  error "Backtrace:\n" + e.backtrace.join("\n")
end