Class: PrometheusConfigBuilder::ScrapeConfigECS

Inherits:
Object
  • Object
show all
Extended by:
PrometheusConfigBuilderLogger
Defined in:
lib/prometheus-config-builder/scrape_ecs.rb

Constant Summary collapse

@@ecs =

Map of different roles to ecs clients. use get_client() to obtain them

{}

Class Method Summary collapse

Methods included from PrometheusConfigBuilderLogger

logger, logger, logger=

Class Method Details

.get_client(region, assume_role = nil) ⇒ Object 



13
14
15
16
17
18
19
20
21
22
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
 
# File 'lib/prometheus-config-builder/scrape_ecs.rb', line 13

def self.get_client(region, assume_role=nil)

  name = region + "/default"

  if assume_role == nil
    client = @@ecs[name]
    if client == nil
      logger.info("Creating default client: #{name} with region: #{region}")
      client = Aws::ECS::Client.new({
        region: region,
        instance_profile_credentials_retries: 4,
        instance_profile_credentials_timeout: 2,
      })
      @@ecs[name] = client
    end
    return client
  end

  role_arn = assume_role["role_arn"]
  if !role_arn
    raise ArgumentError.new("role_arn not found inside assume_role config structure")
  end
  role_session_name = (assume_role["role_session_name"] or "prometheus-scrape-ecs")

  name = region + "/" + role_arn + "/" + role_session_name
  client = @@ecs[name]
  if client == nil
    logger.info("Using sts:AssumeRole on #{role_arn} as client #{name}")
    role_credentials = Aws::AssumeRoleCredentials.new(
      client: Aws::STS::Client.new({
        region: region
      }),
      role_arn: role_arn,
      role_session_name: role_session_name
    )

    client = Aws::ECS::Client.new(credentials: role_credentials)
    @@ecs[name] = client
  end
  return client
end

.get_task_endpoints(client, cluster, tasks, common_labels, metrics_port) ⇒ Object 



158
159
160
161
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
 
# File 'lib/prometheus-config-builder/scrape_ecs.rb', line 158

def self.get_task_endpoints(client, cluster, tasks, common_labels, metrics_port)
  endpoints = []
  if tasks.length == 0
    return []
  end

  task_chunk = tasks.pop(100)
  loop do
    begin
      options = {
        cluster: cluster,
        tasks: task_chunk,
      }
      result = client.describe_tasks(options)
      if result && result.tasks
        result.tasks.each do |task|
          # FIXME: This assumes somewhat on the ip structure.
          if task.containers[0].network_interfaces && task.containers[0].network_interfaces.length > 0
            ip = task.containers[0].network_interfaces[0].private_ipv_4_address
          else
            logger.warn("WARNING: Unable to obtain ipv4 address from task #{task}")
            next
          end

          if metrics_port
            ip = ip + ":" + metrics_port.to_s
          end
          labels = common_labels.clone
          labels["ecs_arn"] = task.task_arn
          endpoints << {
            "targets" => [ip],
            "labels" => labels
          }
        end
      end
    rescue Aws::ECS::Errors::ServiceError => e
      logger.warn("Error listing ecs tasks for cluster #{cluster}: #{e}\n")
      sleep(1)
      next
    end
    break if tasks.length == 0
    task_chunk = tasks.pop(100)
  end
  return endpoints
end

.get_tasks(client, cluster, service_name) ⇒ Object 



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
 
# File 'lib/prometheus-config-builder/scrape_ecs.rb', line 127

def self.get_tasks(client, cluster, service_name)
  tasks = []
  last_result = Aws::ECS::Types::ListTasksResponse.new
  last_result.next_token = nil
  retries = 3
  loop do
    begin
      options = {
        cluster: cluster,
        service_name: service_name,
        next_token: last_result.next_token
      }
      last_result = client.list_tasks(options)
      if last_result && last_result.task_arns
        tasks.push(*last_result.task_arns)
      end
    rescue Aws::ECS::Errors::ServiceError => e
      logger.warn("Error listing ecs tasks for service #{service_name} in cluster #{cluster}: #{e}")
      retries -= 1
      if retries < 0
        raise IOError.new("Unable to get ecs tasks for service #{service_name} in cluster #{cluster}: #{e}")
      end
      sleep(1)
      next
    end
    break if !last_result.next_token
  end

  return tasks
end

.handle(basename, config, dst_prefix) ⇒ Object 



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
 
# File 'lib/prometheus-config-builder/scrape_ecs.rb', line 55

def self.handle(basename, config, dst_prefix)

  x = $VERBOSE
  $VERBOSE = nil
  require 'aws-sdk'

  if !config["region"]
    logger.warn("File #{basename}: the scrape_configs of type:ecs-tasks doesn't have \"region\" field set. Ignoring!")
    return nil
  end

  region = (config["region"] or "us-east-1")
  assume_role = config["assume_role"]

  retries = 5
  begin
    client = get_client(region, assume_role)

    $VERBOSE = x
    tasks = get_tasks(client, config["cluster"], config["service"])
    endpoints = get_task_endpoints(client, config["cluster"], tasks, config["labels"], config["metrics_port"])

    if !config["job_name"]
      logger.warn("File #{basename}: the scrape_configs of type:ecs-tasks doesn't have \"job_name\" field set. Ignoring!")
      return nil
    end

  rescue Aws::Errors::MissingCredentialsError => e
    logger.warn("Got a Aws::Errors::MissingCredentialsError: #{e}")
    if retries > 0
      retries = retries - 1
      logger.warn("Will try to retry")
      sleep(5)
      retry
    end
    raise e
  rescue ArgumentError => e
    logger.warn("File #{basename}: the scrape_configs of type:ecs-tasks had invalid arguments: #{e}")
    return nil
  end


  # Prometheus might notice that the scrape file is changed in the middle of the write. That's why we'll
  # first write the new contents into a temp file and atomically replace the file with the temp file.
  file = File.expand_path(dst_prefix + "_" + config["job_name"] + ".json")
  tmpfile = Tempfile.new().path
  File.open(tmpfile, "w") do |f|
    f.write(endpoints.to_json)
  end
  File.rename(tmpfile, file)


  # Make copy of the settings and remove our custom properties from it.
  # The rest user can set just as he wants according to the Prometheus schema.
  # See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config
  settings = config.clone
  settings.delete("type")
  settings.delete("assume_role")
  settings.delete("cluster")
  settings.delete("service")
  settings.delete("metrics_port")
  settings.delete("labels")
  settings.delete("region")
  settings["file_sd_configs"] = [
    "files" => [
      file
    ]
  ]

  return settings
end