ring-sqa 0.0.18 → 0.0.19

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: daa3666fef098532d9a3d8758e1f106ce9d0e63e
-  data.tar.gz: 60959c28ce98980b6045b9708a2e9e348c4504f8
+  metadata.gz: 3118fdb4ee6c8ef1701ba4290bdee0c82bcb07db
+  data.tar.gz: 5dc539f039179912322562d3354ceec8b55b99b5
 SHA512:
-  metadata.gz: 8d81bd070959777b765c7654e8912f624cc2293f9ebde5edee4305a6b03545e49d59e3366dd7d2153dcefac13886415bd29fd5a0e9486fca04c1436e3816bb2d
-  data.tar.gz: 9d9b16dbfe7f85ed3420e0f9af3a095590fc1191bd178ccb6d785e35ee6ef03fdcd46c969689e99c32599f02b56198e9319566d14ec1d78e369d5c5df7217b59
+  metadata.gz: bcdfe20771cb15f22135512d9861a7c56b3eafbd185e42f3cdfefa3ed7950f104e3bcda33d3026713a90d5250efba22e3449d02fb1f47cd6023f6e177bcb070e
+  data.tar.gz: eee37b25fb65db3036e90ebc5327c093bbce7f2c67a453e16a8ebc42f5e59e0d1d661b108091812e5e80bb35c3f2d93248747235d0f4bd4f22ebd4f737dedaec

data/lib/ring/sqa/alarm.rb

@@ -30,8 +30,7 @@ class SQA
 
     private
 
-    def initialize database
-      @db       = database
+    def initialize
       @methods  = []
       @methods  << Email.new   if CFG.email.to?
       @methods  << UDP2IRC.new if CFG.irc.password?
@@ -42,19 +41,18 @@ class SQA
     def compose_message alarm_buffer
       exceeding_nodes = alarm_buffer.exceeding_nodes
       msg = {short: "#{@hostname}: raising alarm - #{exceeding_nodes.size} new nodes down"}
-      nodes = NodesJSON.new
+      nodes_json = NodesJSON.new
+      exceeding_nodes = exceeding_nodes.map { |node| nodes_json.get node }
 
       nodes_list = ''
-      exceeding_nodes.sort!.each do |node|
-        json = nodes.get node
-        nodes_list << "- %-30s %14s AS%5s %2s\n" % [json['hostname'], node, json['asn'], json['countrycode']]
+      exceeding_nodes.sort_by{ |node| node[:cc] }.each do |node|
+        nodes_list << "- %-35s %15s  AS%-6s  %2s\n" % [node[:name], node[:ip], node[:as], node[:cc]]
       end
 
       mtr_list = ''
       exceeding_nodes.sample(3).each do |node|
-        json = nodes.get node
-        mtr_list << "%-30s AS%5s (%2s)\n" % [json['hostname'], json['asn'], json['countrycode']]
-        mtr_list << MTR.run(node)
+        mtr_list << "%-35s AS%-6s (%2s)\n" % [node[:name], node[:as], node[:cc]]
+        mtr_list << MTR.run(node[:ip])
         mtr_list << "\n"
       end
 
@@ -62,35 +60,36 @@ class SQA
       time = alarm_buffer.array.size-1
       alarm_buffer.array.each do |ary|
         buffer_list << "%2s min ago %3s measurements failed" % [time, ary.size/2]
-        type = time.to_i < 3 ? " (raised alarm)\n" : " (baseline)\n"
-        buffer_list << type
+        buffer_list << (time.to_i < 3 ? " (raised alarm)\n" : " (baseline)\n")
         time -= 1
       end
 
       msg[:long] = <<EOF
-This is an automated alert from the distributed partial outage 
+Regarding: #{hostname}
+
+This is an automated alert from the distributed partial outage
 monitoring system "RING SQA".
 
-At #{Time.now.utc} the following measurements were analysed 
-as indicating that there is a high probability your NLNOG RING node 
-cannot reach the entire internet. Possible causes could be an outage 
+At #{Time.now.utc} the following measurements were analysed
+as indicating that there is a high probability your NLNOG RING node
+cannot reach the entire internet. Possible causes could be an outage
 in your upstream's or peer's network.
 
-The following nodes previously were reachable, but became unreachable 
+The following #{exceeding_nodes.size} nodes previously were reachable, but became unreachable
 over the course of the last 3 minutes:
 
 #{nodes_list}
 
-As a debug starting point 3 traceroutes were launched right after 
+As a debug starting point 3 traceroutes were launched right after
 detecting the event, they might assist in pinpointing what broke:
 
 #{mtr_list}
 
-An alarm is raised under the following conditions: every 30 seconds 
-your node pings all other nodes. The amount of nodes that cannot be 
-reached is stored in a circular buffer, with each element representing 
-a minute of measurements. In the event that the last three minutes are 
-#{Ring::SQA::CFG.analyzer.tolerance} above the median of the previous 27 measurement slots, a partial 
+An alarm is raised under the following conditions: every 30 seconds
+your node pings all other nodes. The amount of nodes that cannot be
+reached is stored in a circular buffer, with each element representing
+a minute of measurements. In the event that the last three minutes are
+#{Ring::SQA::CFG.analyzer.tolerance} above the median of the previous 27 measurement slots, a partial
 outage is assumed. The ring buffer's output is as following:
 
 #{buffer_list}

data/lib/ring/sqa/analyzer.rb

@@ -30,39 +30,38 @@ class SQA
     def initialize database, nodes
       @db         = database
       @nodes      = nodes
-      @alarm      = Alarm.new @db
-      @buffer     = AnalyzeBuffer.new
+      @alarm      = Alarm.new
+      @buffer     = AnalyzeBuffer.new @nodes.list.size
       @db_id_seen = 0
     end
   end
 
   class AnalyzeBuffer
     attr_reader :array
-    def initialize max_size=30
-      @max_size = max_size
-      init_nodes = Array.new 99, ''
-      @array = Array.new max_size, init_nodes
+    def initialize nodes_count, max_size=30, median_of=27
+      @max_size  = max_size
+      @median_of = median_of
+      init_nodes = Array.new nodes_count * 2, ''
+      @array     = Array.new max_size, init_nodes
     end
     def push e
       @array.shift
       @array.push e
     end
-    def median of_first=27
-      of_first = of_first-1
-      middle   = of_first/2
-      node_count[0..of_first].sort[middle]
+    def median
+      last   = @median_of-1
+      node_count[0..last].sort[last/2]
     end
-    def exceed_median? last=3, tolerance=CFG.analyzer.tolerance
-      first = @max_size-last
+    def exceed_median? tolerance=CFG.analyzer.tolerance
       violate = (median+1)*tolerance
-      node_count[first..-1].all? { |e| e > violate }
+      node_count[@median_of..-1].all? { |e| e > violate }
     end
     def node_count
       @array.map { |nodes| nodes.size }
     end
     def exceeding_nodes
-      exceed = @array[27] & @array[28] & @array[29]
-      exceed - @array[0..26].flatten.uniq
+      exceed = @array[@median_of..-1].inject :&
+      exceed - @array[0..@median_of-1].flatten.uniq
     end
   end
 

data/lib/ring/sqa/nodes.rb

@@ -24,6 +24,7 @@ class SQA
     end
 
     def get_list
+      Log.info "loading #{FILE}"
       list = []
       File.read(FILE).lines.each do |line|
         entry = line.split(/\s+/)

data/lib/ring/sqa/nodes_json.rb

@@ -5,7 +5,13 @@ class SQA
 
   class NodesJSON
     def get node
-      (@nodes[node] or {})
+      json = (@nodes[node] or {})
+      {
+        name: json['hostname'],
+        ip:   node,
+        as:   json['asn'],
+        cc:   json['countrycode'],
+      }
     rescue
       {}
     end

data/ring-sqa.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name              = 'ring-sqa'
-  s.version           = '0.0.18'
+  s.version           = '0.0.19'
   s.licenses          = %w( Apache-2.0 )
   s.platform          = Gem::Platform::RUBY
   s.authors           = [ 'Saku Ytti', 'Job Snijders' ]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ring-sqa
 version: !ruby/object:Gem::Version
-  version: 0.0.18
+  version: 0.0.19
 platform: ruby
 authors:
 - Saku Ytti
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-07-24 00:00:00.000000000 Z
+date: 2014-07-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: slop