interferon 0.0.12 → 0.0.19
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +13 -5
- data/interferon.gemspec +6 -2
- data/lib/interferon/alert.rb +8 -0
- data/lib/interferon/alert_dsl.rb +10 -0
- data/lib/interferon/destinations/datadog.rb +127 -55
- data/lib/interferon/group_sources/filesystem.rb +18 -2
- data/lib/interferon/version.rb +1 -1
- data/lib/interferon/work_hours_helper.rb +21 -0
- data/lib/interferon.rb +134 -45
- data/spec/helpers/dsl_helper.rb +10 -1
- data/spec/helpers/mock_alert.rb +1 -0
- data/spec/lib/interferon/group_sources/filesystem_spec.rb +57 -0
- data/spec/lib/interferon_spec.rb +190 -43
- data/spec/lib/work_hours_helper_spec.rb +41 -0
- metadata +106 -25
data/lib/interferon.rb
CHANGED
@@ -9,6 +9,7 @@ require 'interferon/alert_dsl'
|
|
9
9
|
#require 'pry' #uncomment if you're debugging
|
10
10
|
require 'erb'
|
11
11
|
require 'ostruct'
|
12
|
+
require 'parallel'
|
12
13
|
require 'set'
|
13
14
|
require 'yaml'
|
14
15
|
|
@@ -23,11 +24,14 @@ module Interferon
|
|
23
24
|
# groups_sources is a hash from type => options for each group source
|
24
25
|
# host_sources is a hash from type => options for each host source
|
25
26
|
# destinations is a similiar hash from type => options for each alerter
|
26
|
-
def initialize(alerts_repo_path, groups_sources, host_sources, destinations
|
27
|
+
def initialize(alerts_repo_path, groups_sources, host_sources, destinations,
|
28
|
+
dry_run=false, processes=nil)
|
27
29
|
@alerts_repo_path = alerts_repo_path
|
28
30
|
@groups_sources = groups_sources
|
29
31
|
@host_sources = host_sources
|
30
32
|
@destinations = destinations
|
33
|
+
@dry_run = dry_run
|
34
|
+
@processes = processes
|
31
35
|
@request_shutdown = false
|
32
36
|
end
|
33
37
|
|
@@ -36,7 +40,8 @@ module Interferon
|
|
36
40
|
log.info "SIGTERM received. shutting down gracefully..."
|
37
41
|
@request_shutdown = true
|
38
42
|
end
|
39
|
-
|
43
|
+
@dry_run = dry_run
|
44
|
+
run_desc = @dry_run ? 'dry run' : 'run'
|
40
45
|
log.info "beginning alerts #{run_desc}"
|
41
46
|
|
42
47
|
alerts = read_alerts
|
@@ -45,9 +50,12 @@ module Interferon
|
|
45
50
|
|
46
51
|
@destinations.each do |dest|
|
47
52
|
dest['options'] ||= {}
|
53
|
+
if @dry_run
|
54
|
+
dest['options']['dry_run'] = true
|
55
|
+
end
|
48
56
|
end
|
49
57
|
|
50
|
-
update_alerts(@destinations, hosts, alerts, groups
|
58
|
+
update_alerts(@destinations, hosts, alerts, groups)
|
51
59
|
|
52
60
|
if @request_shutdown
|
53
61
|
log.info "interferon #{run_desc} shut down by SIGTERM"
|
@@ -133,23 +141,23 @@ module Interferon
|
|
133
141
|
return hosts
|
134
142
|
end
|
135
143
|
|
136
|
-
def update_alerts(destinations, hosts, alerts, groups
|
144
|
+
def update_alerts(destinations, hosts, alerts, groups)
|
137
145
|
loader = DestinationsLoader.new([@alerts_repo_path])
|
138
146
|
loader.get_all(destinations).each do |dest|
|
139
147
|
break if @request_shutdown
|
140
148
|
log.info "updating alerts on #{dest.class.name}"
|
141
|
-
update_alerts_on_destination(dest, hosts, alerts, groups
|
149
|
+
update_alerts_on_destination(dest, hosts, alerts, groups)
|
142
150
|
end
|
143
151
|
end
|
144
152
|
|
145
|
-
def update_alerts_on_destination(dest, hosts, alerts, groups
|
153
|
+
def update_alerts_on_destination(dest, hosts, alerts, groups)
|
146
154
|
# track some counters/stats per destination
|
147
155
|
start_time = Time.new.to_f
|
148
156
|
|
149
157
|
# get already-defined alerts
|
150
|
-
existing_alerts = dest.existing_alerts
|
158
|
+
existing_alerts = dest.existing_alerts
|
151
159
|
|
152
|
-
if dry_run
|
160
|
+
if @dry_run
|
153
161
|
do_dry_run_update(dest, hosts, alerts, existing_alerts, groups)
|
154
162
|
else
|
155
163
|
do_regular_update(dest, hosts, alerts, existing_alerts, groups)
|
@@ -159,7 +167,7 @@ module Interferon
|
|
159
167
|
# run time summary
|
160
168
|
run_time = Time.new.to_f - start_time
|
161
169
|
statsd.histogram(
|
162
|
-
dry_run ? 'destinations.run_time.dry_run' : 'destinations.run_time',
|
170
|
+
@dry_run ? 'destinations.run_time.dry_run' : 'destinations.run_time',
|
163
171
|
run_time,
|
164
172
|
:tags => ["destination:#{dest.class.name}"])
|
165
173
|
log.info "#{dest.class.name} : run completed in %.2f seconds" % (run_time)
|
@@ -168,46 +176,103 @@ module Interferon
|
|
168
176
|
dest.report_stats
|
169
177
|
end
|
170
178
|
|
171
|
-
if dry_run && !dest.api_errors.empty?
|
179
|
+
if @dry_run && !dest.api_errors.empty?
|
172
180
|
raise dest.api_errors.to_s
|
173
181
|
end
|
174
182
|
end
|
175
183
|
|
176
184
|
def do_dry_run_update(dest, hosts, alerts, existing_alerts, groups)
|
177
|
-
|
185
|
+
# Track these to clean up dry-run alerts from previous runs
|
186
|
+
existing_dry_run_alerts = []
|
187
|
+
existing_alerts.each do |name, alert|
|
188
|
+
if name.start_with?(DRY_RUN_ALERTS_NAME_PREFIX)
|
189
|
+
existing_dry_run_alerts << [alert['name'], [alert['id']]]
|
190
|
+
existing_alerts.delete(name)
|
191
|
+
end
|
192
|
+
end
|
193
|
+
|
178
194
|
alerts_queue = build_alerts_queue(hosts, alerts, groups)
|
179
|
-
alerts_queue.reject
|
180
|
-
|
181
|
-
|
182
|
-
|
195
|
+
updates_queue = alerts_queue.reject do |name, alert_people_pair|
|
196
|
+
!Interferon::need_update(dest, alert_people_pair, existing_alerts)
|
197
|
+
end
|
198
|
+
|
199
|
+
# Add dry-run prefix to alerts and delete id to avoid impacting real alerts
|
200
|
+
existing_alerts.keys.each do |name|
|
201
|
+
existing_alert = existing_alerts[name]
|
202
|
+
dry_run_alert_name = DRY_RUN_ALERTS_NAME_PREFIX + name
|
203
|
+
existing_alert['name'] = dry_run_alert_name
|
204
|
+
existing_alert['id'] = [nil]
|
205
|
+
existing_alerts[dry_run_alert_name] = existing_alerts.delete(name)
|
206
|
+
end
|
207
|
+
|
208
|
+
# Build new queue with dry-run prefixes and ensure they are silenced
|
209
|
+
alerts_queue.each do |name, alert_people_pair|
|
210
|
+
alert = alert_people_pair[0]
|
211
|
+
dry_run_alert_name = DRY_RUN_ALERTS_NAME_PREFIX + alert['name']
|
212
|
+
alert.change_name(dry_run_alert_name)
|
213
|
+
alert.silence
|
214
|
+
end
|
215
|
+
|
216
|
+
# Create alerts in destination
|
217
|
+
created_alerts = create_alerts(dest, updates_queue)
|
218
|
+
|
219
|
+
# Existing alerts are pruned until all that remains are alerts that aren't being generated anymore
|
220
|
+
to_remove = existing_alerts.dup
|
221
|
+
alerts_queue.each do |name, alert_people_pair|
|
222
|
+
alert = alert_people_pair[0]
|
223
|
+
old_alerts = to_remove[alert['name']]
|
224
|
+
|
225
|
+
if !old_alerts.nil?
|
226
|
+
if old_alerts['id'].length == 1
|
227
|
+
to_remove.delete(alert['name'])
|
228
|
+
else
|
229
|
+
old_alerts['id'] = old_alerts['id'].drop(1)
|
230
|
+
end
|
231
|
+
end
|
183
232
|
end
|
184
233
|
|
185
|
-
#
|
186
|
-
|
187
|
-
created_alerts_ids = created_alerts_key_ids.map{|a| a[1]}
|
188
|
-
to_remove_ids = to_remove.empty? ? [] : to_remove.map{|a| a['id']}
|
189
|
-
# remove existing alerts that shouldn't exist
|
190
|
-
(created_alerts_ids + to_remove_ids).each do |id|
|
234
|
+
# Clean up alerts not longer being generated
|
235
|
+
to_remove.each do |name, alert|
|
191
236
|
break if @request_shutdown
|
192
|
-
dest.
|
237
|
+
dest.remove_alert(alert)
|
238
|
+
end
|
239
|
+
|
240
|
+
# Clean up dry-run created alerts
|
241
|
+
(created_alerts + existing_dry_run_alerts).each do |alert_id_pair|
|
242
|
+
alert_ids = alert_id_pair[1]
|
243
|
+
alert_ids.each do |alert_id|
|
244
|
+
dest.remove_alert_by_id(alert_id)
|
245
|
+
end
|
193
246
|
end
|
247
|
+
|
194
248
|
end
|
195
249
|
|
196
250
|
def do_regular_update(dest, hosts, alerts, existing_alerts, groups)
|
197
|
-
existing_alerts.each{ |key, existing_alert| existing_alert['still_exists'] = false }
|
198
|
-
|
199
251
|
alerts_queue = build_alerts_queue(hosts, alerts, groups)
|
252
|
+
updates_queue = alerts_queue.reject do |name, alert_people_pair|
|
253
|
+
!Interferon::need_update(dest, alert_people_pair, existing_alerts)
|
254
|
+
end
|
200
255
|
|
201
|
-
#
|
202
|
-
|
203
|
-
|
204
|
-
|
205
|
-
|
256
|
+
# Create alerts in destination
|
257
|
+
create_alerts(dest, updates_queue)
|
258
|
+
|
259
|
+
# Existing alerts are pruned until all that remains are alerts that aren't being generated anymore
|
260
|
+
to_remove = existing_alerts.dup
|
261
|
+
alerts_queue.each do |name, alert_people_pair|
|
262
|
+
alert = alert_people_pair[0]
|
263
|
+
old_alerts = to_remove[alert['name']]
|
264
|
+
|
265
|
+
if !old_alerts.nil?
|
266
|
+
if old_alerts['id'].length == 1
|
267
|
+
to_remove.delete(alert['name'])
|
268
|
+
else
|
269
|
+
old_alerts['id'] = old_alerts['id'].drop(1)
|
270
|
+
end
|
271
|
+
end
|
206
272
|
end
|
207
273
|
|
208
|
-
#
|
209
|
-
|
210
|
-
to_delete.each do |key, alert|
|
274
|
+
# Clean up alerts not longer being generated
|
275
|
+
to_remove.each do |name, alert|
|
211
276
|
break if @request_shutdown
|
212
277
|
dest.remove_alert(alert)
|
213
278
|
end
|
@@ -237,10 +302,11 @@ module Interferon
|
|
237
302
|
end
|
238
303
|
|
239
304
|
def build_alerts_queue(hosts, alerts, groups)
|
305
|
+
alerts_queue = {}
|
240
306
|
# create or update alerts; mark when we've done that
|
241
|
-
|
242
|
-
alerts.each do |alert|
|
307
|
+
result = Parallel.map(alerts, in_processes: @processes) do |alert|
|
243
308
|
break if @request_shutdown
|
309
|
+
alerts_generated = {}
|
244
310
|
counters = {
|
245
311
|
:errors => 0,
|
246
312
|
:evals => 0,
|
@@ -268,7 +334,7 @@ module Interferon
|
|
268
334
|
|
269
335
|
counters[:applies] += 1
|
270
336
|
# don't define alerts twice
|
271
|
-
next if
|
337
|
+
next if alerts_generated.key?(alert[:name])
|
272
338
|
|
273
339
|
# figure out who to notify
|
274
340
|
people = Set.new(alert[:notify][:people])
|
@@ -277,7 +343,7 @@ module Interferon
|
|
277
343
|
end
|
278
344
|
|
279
345
|
# queue the alert up for creation; we clone the alert to save the current state
|
280
|
-
|
346
|
+
alerts_generated[alert[:name]] = [alert.clone, people]
|
281
347
|
end
|
282
348
|
|
283
349
|
# log some of the counters
|
@@ -289,7 +355,7 @@ module Interferon
|
|
289
355
|
end
|
290
356
|
|
291
357
|
# did the alert fail to evaluate on all hosts?
|
292
|
-
if counters[:errors] == counters[:hosts]
|
358
|
+
if counters[:errors] == counters[:hosts] && !last_eval_error.nil?
|
293
359
|
log.error "alert #{alert} failed to evaluate in the context of all hosts!"
|
294
360
|
log.error "last error on alert #{alert}: #{last_eval_error}"
|
295
361
|
|
@@ -306,25 +372,48 @@ module Interferon
|
|
306
372
|
else
|
307
373
|
statsd.gauge('alerts.evaluate.never_applies', 0, :tags => ["alert:#{alert}"])
|
308
374
|
end
|
375
|
+
alerts_generated
|
376
|
+
end
|
377
|
+
|
378
|
+
result.each do |alerts_generated|
|
379
|
+
alerts_queue.merge! alerts_generated
|
309
380
|
end
|
310
381
|
alerts_queue
|
311
382
|
end
|
312
383
|
|
313
|
-
def self.
|
384
|
+
def self.need_update(dest, alert_people_pair, existing_alerts_from_api)
|
385
|
+
alert = alert_people_pair[0]
|
314
386
|
existing = existing_alerts_from_api[alert['name']]
|
315
387
|
if existing.nil?
|
316
388
|
true
|
317
389
|
else
|
318
|
-
!
|
390
|
+
!same_alerts(dest, alert_people_pair, existing)
|
319
391
|
end
|
320
392
|
end
|
321
393
|
|
322
|
-
def self.
|
323
|
-
|
324
|
-
|
325
|
-
|
326
|
-
|
327
|
-
|
394
|
+
def self.same_alerts(dest, alert_people_pair, alert_api_json)
|
395
|
+
alert, people = alert_people_pair
|
396
|
+
|
397
|
+
prev_alert = {
|
398
|
+
:query => alert_api_json['query'].strip,
|
399
|
+
:message => alert_api_json['message'].strip,
|
400
|
+
:notify_no_data => alert_api_json['notify_no_data'],
|
401
|
+
:silenced => alert_api_json['silenced'],
|
402
|
+
:timeout => alert_api_json['timeout_h'],
|
403
|
+
:no_data_timeframe => alert_api_json['no_data_timeframe']
|
404
|
+
}
|
405
|
+
|
406
|
+
new_alert = {
|
407
|
+
:query => alert['metric']['datadog_query'].strip,
|
408
|
+
:message => dest.generate_message(alert['message'], people).strip,
|
409
|
+
:notify_no_data => alert['notify_no_data'],
|
410
|
+
:silenced => alert['silenced'] || alert['silenced_until'] > Time.now,
|
411
|
+
:timeout => alert['timeout'] ? [1, alert['timeout'].to_i / 3600].max : nil,
|
412
|
+
:no_data_timeframe => alert['no_data_timeframe'] || nil
|
413
|
+
}
|
414
|
+
|
415
|
+
prev_alert == new_alert
|
328
416
|
end
|
417
|
+
|
329
418
|
end
|
330
419
|
end
|
data/spec/helpers/dsl_helper.rb
CHANGED
@@ -6,7 +6,8 @@ module Interferon
|
|
6
6
|
def get_or_set(field, val, block, default)
|
7
7
|
@hash ||= Hash.new
|
8
8
|
if val.nil?
|
9
|
-
|
9
|
+
f = @hash[field]
|
10
|
+
f.nil? ? default : f
|
10
11
|
else
|
11
12
|
@hash[field] = val
|
12
13
|
end
|
@@ -27,6 +28,14 @@ module Interferon
|
|
27
28
|
def id(v = nil, &block)
|
28
29
|
get_or_set(:@id, v, block, '')
|
29
30
|
end
|
31
|
+
|
32
|
+
def silenced(v = nil, &block)
|
33
|
+
get_or_set(:@silenced, v, block, false)
|
34
|
+
end
|
35
|
+
|
36
|
+
def silenced_until(v = nil, &block)
|
37
|
+
get_or_set(:@silenced_until, v && Time.parse(v), block, Time.at(0))
|
38
|
+
end
|
30
39
|
end
|
31
40
|
|
32
41
|
class MockNotifyDSL < NotifyDSL
|
data/spec/helpers/mock_alert.rb
CHANGED
@@ -0,0 +1,57 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
require 'interferon/group_sources/filesystem'
|
3
|
+
|
4
|
+
describe Interferon::GroupSources::Filesystem do
|
5
|
+
let (:fs_loader) { Interferon::GroupSources::Filesystem.new({'paths' => ['/tmp']}) }
|
6
|
+
|
7
|
+
describe 'list_groups' do
|
8
|
+
context "with basic groups" do
|
9
|
+
before do
|
10
|
+
group_a = double()
|
11
|
+
expect(File).to receive(:read).with('group_a.yaml').and_return('group_a_text')
|
12
|
+
expect(Psych).to receive(:parse).and_return(group_a)
|
13
|
+
expect(group_a).to receive(:to_ruby).and_return({'name' => 'group_a',
|
14
|
+
'people' => ['Alice', 'Bob']})
|
15
|
+
|
16
|
+
group_b = double()
|
17
|
+
expect(File).to receive(:read).with('group_b.yaml').and_return('group_b_text')
|
18
|
+
expect(Psych).to receive(:parse).and_return(group_b)
|
19
|
+
expect(group_b).to receive(:to_ruby).and_return({'name' => 'group_b',
|
20
|
+
'people' => ['Carol', 'Dave']})
|
21
|
+
end
|
22
|
+
|
23
|
+
it 'loads groups defined by YAML' do
|
24
|
+
expect(Dir).to receive(:glob).and_return(['group_a.yaml', 'group_b.yaml'].each)
|
25
|
+
|
26
|
+
groups = fs_loader.list_groups()
|
27
|
+
expect(groups).to eq({'group_a' => ['Alice', 'Bob'], 'group_b' => ['Carol', 'Dave']})
|
28
|
+
end
|
29
|
+
|
30
|
+
it 'allows groups to be aliased in YAML' do
|
31
|
+
expect(Dir).to receive(:glob).and_return(['group_a.yaml', 'group_b.yaml', 'group_c.yaml'].each)
|
32
|
+
group_c = double()
|
33
|
+
expect(File).to receive(:read).with('group_c.yaml').and_return('group_c_text')
|
34
|
+
expect(Psych).to receive(:parse).and_return(group_c)
|
35
|
+
expect(group_c).to receive(:to_ruby).and_return({'name' => 'group_c', 'alias_for' => 'group_b'})
|
36
|
+
|
37
|
+
groups = fs_loader.list_groups()
|
38
|
+
expect(groups).to eq({'group_a' => ['Alice', 'Bob'],
|
39
|
+
'group_b' => ['Carol', 'Dave'],
|
40
|
+
'group_c' => ['Carol', 'Dave']})
|
41
|
+
end
|
42
|
+
|
43
|
+
it 'skips bad aliases in YAML' do
|
44
|
+
expect(Dir).to receive(:glob).and_return(['group_a.yaml', 'group_b.yaml', 'group_c.yaml'].each)
|
45
|
+
group_c = double()
|
46
|
+
expect(File).to receive(:read).with('group_c.yaml').and_return('group_c_text')
|
47
|
+
expect(Psych).to receive(:parse).and_return(group_c)
|
48
|
+
expect(group_c).to receive(:to_ruby).and_return({'name' => 'group_c', 'alias_for' => 'group_d'})
|
49
|
+
|
50
|
+
groups = fs_loader.list_groups()
|
51
|
+
expect(groups).to eq({'group_a' => ['Alice', 'Bob'],
|
52
|
+
'group_b' => ['Carol', 'Dave']})
|
53
|
+
end
|
54
|
+
end
|
55
|
+
end
|
56
|
+
|
57
|
+
end
|