Class: Bosh::Director::ProblemScanner

Inherits:
Object
  • Object
show all
Defined in:
lib/bosh/director/problem_scanner.rb

Constant Summary collapse

AGENT_TIMEOUT =

seconds

10

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(deployment) ⇒ ProblemScanner

Returns a new instance of ProblemScanner.

Parameters:

  • deployment_name (String) —

    Deployment name



13
14
15
16
17
18
19
20
21
22
23
 
# File 'lib/bosh/director/problem_scanner.rb', line 13

def initialize(deployment)
  @deployment = deployment
  @instance_manager = Api::InstanceManager.new

  @problem_lock = Mutex.new
  @agent_disks = {}

  #temp
  @event_log = Config.event_log
  @logger = Config.logger
end

Instance Attribute Details

#event_log ⇒ Object (readonly)

Returns the value of attribute event_log.



8
9
10
 
# File 'lib/bosh/director/problem_scanner.rb', line 8

def event_log
  @event_log
end

#logger ⇒ Object (readonly)

Returns the value of attribute logger.



8
9
10
 
# File 'lib/bosh/director/problem_scanner.rb', line 8

def logger
  @logger
end

Instance Method Details

#begin_stage(stage_name, n_steps) ⇒ Object 



25
26
27
28
 
# File 'lib/bosh/director/problem_scanner.rb', line 25

def begin_stage(stage_name, n_steps)
  event_log.begin_stage(stage_name, n_steps)
  logger.info(stage_name)
end

#problem_found(type, resource, data = {}) ⇒ Object 



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
 
# File 'lib/bosh/director/problem_scanner.rb', line 191

def problem_found(type, resource, data = {})
  @problem_lock.synchronize do
    similar_open_problems = Models::DeploymentProblem.
        filter(:deployment_id => deployment.id, :type => type.to_s,
               :resource_id => resource.id, :state => "open").all

    if similar_open_problems.size > 1
      raise CloudcheckTooManySimilarProblems,
            "More than one problem of type `#{type}' " +
                "exists for resource #{type} #{resource.id}"
    end

    if similar_open_problems.empty?
      problem = Models::DeploymentProblem.
          create(:type => type.to_s, :resource_id => resource.id,
                 :state => "open", :deployment_id => deployment.id,
                 :data => data, :counter => 1)

      logger.info("Created problem #{problem.id} (#{problem.type})")
    else
      # This assumes we are running with deployment lock acquired,
      # so there is no possible update conflict
      problem = similar_open_problems[0]
      problem.data = data
      problem.last_seen_at = Time.now
      problem.counter += 1
      problem.save
      logger.info("Updated problem #{problem.id} (#{problem.type}), " +
                      "count is now #{problem.counter}")
    end
  end
end

#reset(vms = nil) ⇒ Object 



37
38
39
40
41
42
43
44
45
46
47
48
 
# File 'lib/bosh/director/problem_scanner.rb', line 37

def reset(vms=nil)
  if vms
    vms.each do |job, index|
      instance = @instance_manager.find_by_name(@deployment.name, job, index)
      Models::DeploymentProblem.where(deployment: deployment,
                                      :resource_id => instance.vm.id,
                                      :state => "open").update(state: "closed")
    end
  else
    Models::DeploymentProblem.where(state: "open", deployment: deployment).update(state: "closed")
  end
end

#scan_disk(disk) ⇒ Object 



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
 
# File 'lib/bosh/director/problem_scanner.rb', line 104

def scan_disk(disk)
  # inactive disks
  unless disk.active
    logger.info("Found inactive disk: #{disk.id}")
    problem_found(:inactive_disk, disk)
    return :inactive
  end

  disk_cid = disk.disk_cid
  vm_cid = nil

  if disk.instance && disk.instance.vm
    vm_cid = disk.instance.vm.cid
  end

  if vm_cid.nil?
    # With the db dependencies this should not happen.
    logger.warn("Disk #{disk_cid} is not associated to any VM. " +
                    "Skipping scan")
    return :ok
  end

  owner_vms = get_disk_owners(disk_cid) || []
  # active disk is not mounted or mounted more than once -or-
  # the disk is mounted on a vm that is different form the record.
  if owner_vms.size != 1 || owner_vms.first != vm_cid
    logger.info("Found problem in mount info: " +
                    "active disk #{disk_cid} mounted on " +
                    "#{owner_vms.join(', ')}")
    problem_found(:mount_info_mismatch, disk, :owner_vms => owner_vms)
    return :mount_info_mismatch
  end
  :ok
end

#scan_disks ⇒ Object 



50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
 
# File 'lib/bosh/director/problem_scanner.rb', line 50

def scan_disks
  disks = Models::PersistentDisk.eager(:instance).all.select do |disk|
    disk.instance && disk.instance.deployment_id == deployment.id
  end
  results = Hash.new(0)

  begin_stage("Scanning #{disks.size} persistent disks", 2)

  track_and_log("Looking for inactive disks") do
    disks.each do |disk|
      scan_result = scan_disk(disk)
      results[scan_result] += 1
    end
  end

  track_and_log("#{results[:ok]} OK, " +
                    "#{results[:inactive]} inactive, " +
                    "#{results[:mount_info_mismatch]} mount-info mismatch")
end

#scan_vm(vm) ⇒ Object 



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
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
 
# File 'lib/bosh/director/problem_scanner.rb', line 139

def scan_vm(vm)
  agent_options = {
      :timeout => AGENT_TIMEOUT,
      :retry_methods => {:get_state => 0}
  }

  instance = nil
  mounted_disk_cid = nil
  @problem_lock.synchronize do
    instance = vm.instance
    mounted_disk_cid = instance.persistent_disk_cid if instance
  end

  agent = AgentClient.with_defaults(vm.agent_id, agent_options)
  begin
    state = agent.get_state

    # gather mounted disk info. (used by scan_disk)
    begin
      disk_list = agent.list_disk
      mounted_disk_cid = disk_list.first
    rescue Bosh::Director::RpcTimeout => e
      mounted_disk_cid = nil
    rescue RuntimeError => e
      # For old agents that doesn't implement list_disk we assume the disk is mounted
      logger.info("agent.list_disk failed on agent #{vm.agent_id}")
    end
    add_disk_owner(mounted_disk_cid, vm.cid) if mounted_disk_cid

    return :out_of_sync if is_out_of_sync_vm?(vm, instance, state)
    return :unbound if is_unbound_instance_vm?(vm, instance, state)
    :ok
  rescue Bosh::Director::RpcTimeout
    # We add the disk to avoid a duplicate problem when timeouts fetching agent status (unresponsive_agent and
    # mount_info_mismatch)
    add_disk_owner(mounted_disk_cid, vm.cid) if mounted_disk_cid

    begin
      unless cloud.has_vm?(vm.cid)
        logger.info("Missing VM #{vm.cid}")
        problem_found(:missing_vm, vm)
        return :missing
      end
    rescue Bosh::Clouds::NotImplemented
    end

    logger.info("Found unresponsive agent #{vm.agent_id}")
    problem_found(:unresponsive_agent, vm)
    :unresponsive
  end
end

#scan_vms(vms = nil) ⇒ Object 



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
 
# File 'lib/bosh/director/problem_scanner.rb', line 70

def scan_vms(vms=nil)
  if vms
    vm_list = []
    vms.each do |job, index|
      instance = @instance_manager.find_by_name(@deployment.name, job, index)
      vm_list << instance.vm
    end
    vms = vm_list
  else
    vms = Models::Vm.eager(:instance).filter(deployment: deployment).all
  end

  begin_stage("Scanning #{vms.size} VMs", 2)
  results = Hash.new(0)
  lock = Mutex.new

  track_and_log("Checking VM states") do
    ThreadPool.new(:max_threads => Config.max_threads).wrap do |pool|
      vms.each do |vm|
        pool.process do
          scan_result = scan_vm(vm)
          lock.synchronize { results[scan_result] += 1 }
        end
      end
    end
  end

  track_and_log("#{results[:ok]} OK, " +
                    "#{results[:unresponsive]} unresponsive, " +
                    "#{results[:missing]} missing, " +
                    "#{results[:unbound]} unbound, " +
                    "#{results[:out_of_sync]} out of sync")
end

#track_and_log(task, log = true) ⇒ Object 



30
31
32
33
34
35
 
# File 'lib/bosh/director/problem_scanner.rb', line 30

def track_and_log(task, log = true)
  event_log.track(task) do |ticker|
    logger.info(task) if log
    yield ticker if block_given?
  end
end