From f0f0e01562339ed94eddee6eb13e92b8f17b6acc Mon Sep 17 00:00:00 2001 From: Steve Kemp Date: Wed, 14 Nov 2012 16:45:50 +0000 Subject: Moved core of the code into their own libraries. --- bin/custodian-dequeue | 285 +---------------------------------------------- lib/custodian.rb | 218 ++++++++++++++++++++++++++++++++++++ lib/custodian/alerter.rb | 76 +++++++++++++ 3 files changed, 299 insertions(+), 280 deletions(-) create mode 100644 lib/custodian.rb create mode 100644 lib/custodian/alerter.rb diff --git a/bin/custodian-dequeue b/bin/custodian-dequeue index 850e169..85bc6f3 100755 --- a/bin/custodian-dequeue +++ b/bin/custodian-dequeue @@ -43,292 +43,17 @@ # - -require 'beanstalk-client' -require 'getoptlong' -require 'json' -require 'logger' - -require 'mauve/sender' -require 'mauve/proto' - - - # -# Implementation of our protocol tests. +# Standard modules # -require 'custodian/protocol-tests/dns.rb' -require 'custodian/protocol-tests/ftp.rb' -require 'custodian/protocol-tests/http.rb' -require 'custodian/protocol-tests/https.rb' -require 'custodian/protocol-tests/jabber.rb' -require 'custodian/protocol-tests/ldap.rb' -require 'custodian/protocol-tests/ping.rb' -require 'custodian/protocol-tests/rsync.rb' -require 'custodian/protocol-tests/smtp.rb' -require 'custodian/protocol-tests/ssh.rb' - - - - - -# -# This class encapsulates the raising and clearing of alerts via Mauve. -# -class Alert - - attr_reader :details - - def initialize( test_details ) - @details = test_details - end - - - # - # Raise the alert. - # - def raise( detail ) - - update = Mauve::Proto::AlertUpdate.new - update.alert = [] - update.source = "custodian" - - # be explicit about raising/clearing - update.replace = false - - alert = Mauve::Proto::Alert.new - - # e.g. ping-example.vm.bytemark.co.uk - # e.g. http-http://example.com/page1 - alert.id = "#{@details['test_type']}-#{@details['target_host']}" - - alert.subject = @details['target_host'] - alert.summary = @details['test_alert'] - alert.detail = "The #{@details['test_type']} test failed against #{@details['target_host']}: #{detail}" - alert.raise_time = Time.now.to_i - update.alert << alert - - Mauve::Sender.new("alert.bytemark.co.uk").send(update) - - end - - # - # Clear the alert. - # - def clear - - update = Mauve::Proto::AlertUpdate.new - update.alert = [] - update.source = "custodian" - - # be explicit about raising/clearing - update.replace = false - - - alert = Mauve::Proto::Alert.new - - # e.g. ping-example.vm.bytemark.co.uk - # e.g. http-http://example.com/page1 - alert.id = "#{@details['test_type']}-#{@details['target_host']}" - - alert.subject = @details['target_host'] - alert.summary = @details['test_alert'] - alert.detail = "The #{@details['test_type']} test succeeded against #{@details['target_host']}" - alert.clear_time = Time.now.to_i - update.alert << alert - - Mauve::Sender.new("alert.bytemark.co.uk").send(update) - end - -end - - +require 'getoptlong' # -# This class contains the code for connecting to a Beanstalk queue, -# fetching tests from it, and executing them +# Our code. # -class Custodian - - # - # The beanstalk queue. - # - attr_reader :queue - - # - # How many times we re-test before we detect a failure - # - attr_reader :retry_count - - # - # The log-file object - # - attr_reader :logger - - # - # Constructor: Connect to the queue - # - def initialize( server, logfile ) - - # Connect to the queue - @queue = Beanstalk::Pool.new([server]) - - # Instantiate the logger. - @logger = Logger.new( logfile, "daily" ) - - if ( ENV['REPEAT'] ) - @retry_count=ENV['REPEAT'].to_i - else - @retry_count=5 - end - - log_message( "We'll run each test #{@retry_count} before alerting failures." ) - end - - - # - # Write the given message to our logfile - and show it to the console - # if we're running with '--verbose' in play - # - def log_message( msg ) - @logger.info( msg ) - puts msg if ( ENV['VERBOSE'] ) - end - - - - # - # Process jobs from the queue - never return. - # - def run! - while( true ) - log_message( "\n" ) - log_message( "\n" ) - log_message( "Waiting for job.." ) - process_single_job() - end - end - - - - # - # Fetch a single job from the queue, and process it. - # - def process_single_job - - begin - job = @queue.reserve() - - log_message( "Job aquired - Job ID : #{job.id}" ) - - - # - # Parse the JSON of the job body. - # - json = job.body - hash = JSON.parse( json ) - hash['verbose'] = 1 if ( ENV['VERBOSE'] ) - - - # - # Output the details. - # - log_message( "Job body contains the following keys & values:") - hash.keys.each do |key| - log_message( " #{key} => #{hash[key]}" ) - end - - - - # - # Did the test succeed? If not count the number of times it failed in - # a row. We'll repeat several times - # - success = false - count = 0 - - # - # As a result of this test we'll either raise/clear with mauve. - # - # This helper will do that job. - # - alert = Alert.new( hash ) - - - # - # Convert the test-type to a class name, to do the protocol test. - # - # Given a test-type "foo" we'll attempt to instantiate a class called FOOTest. - # - test = hash['test_type'] - clazz = test.upcase - clazz = "#{clazz}Test" - - - # - # Create the test object. - # - obj = eval(clazz).new( hash ) - - - # - # Ensure that the object we load implements the two methods - # we expect. - # - if ( ( ! obj.respond_to?( "error") ) || - ( ! obj.respond_to?( "run_test" ) ) ) - puts "Class #{clazz} doesn't implement the full protocol-test API" - end - - - - # - # We'll run no more than MAX times. - # - # We stop the execution on a single success. - # - while ( ( count < @retry_count ) && ( success == false ) ) - - log_message( "Running test - attempt #{count}" ) - - if ( obj.run_test() ) - log_message( "Test succeeed - clearing alert" ) - success = true - alert.clear() - end - count += 1 - end - - # - # If we didn't succeed on any of the attempts raise the alert. - # - if ( ! success ) - - # - # Raise the alert, passing the error message. - # - log_message( "Test failed - alerting with #{obj.error()}" ) - alert.raise( obj.error() ) - end - - rescue => ex - puts "Exception raised processing job: #{ex}" - - ensure - # - # Delete the job - either we received an error, in which case - # we should remove it to avoid picking it up again, or we handled - # it successfully so it should be removed. - # - log_message( "Job ID : #{job.id} - Removed" ) - job.delete if ( job ) - end - end -end - - - +require 'custodian.rb' +require 'custodian/alerter.rb' diff --git a/lib/custodian.rb b/lib/custodian.rb new file mode 100644 index 0000000..79fca72 --- /dev/null +++ b/lib/custodian.rb @@ -0,0 +1,218 @@ + + +# +# Standard modules +# +require 'beanstalk-client' +require 'json' +require 'logger' + + + +# +# Implementation of our protocol tests. +# +require 'custodian/alerter.rb' +require 'custodian/protocol-tests/dns.rb' +require 'custodian/protocol-tests/ftp.rb' +require 'custodian/protocol-tests/http.rb' +require 'custodian/protocol-tests/https.rb' +require 'custodian/protocol-tests/jabber.rb' +require 'custodian/protocol-tests/ldap.rb' +require 'custodian/protocol-tests/ping.rb' +require 'custodian/protocol-tests/rsync.rb' +require 'custodian/protocol-tests/smtp.rb' +require 'custodian/protocol-tests/ssh.rb' + + + + + + + + + +# +# This class contains the code for connecting to a Beanstalk queue, +# fetching tests from it, and executing them +# +class Custodian + + # + # The beanstalk queue. + # + attr_reader :queue + + # + # How many times we re-test before we detect a failure + # + attr_reader :retry_count + + # + # The log-file object + # + attr_reader :logger + + # + # Constructor: Connect to the queue + # + def initialize( server, logfile ) + + # Connect to the queue + @queue = Beanstalk::Pool.new([server]) + + # Instantiate the logger. + @logger = Logger.new( logfile, "daily" ) + + if ( ENV['REPEAT'] ) + @retry_count=ENV['REPEAT'].to_i + else + @retry_count=5 + end + + log_message( "We'll run each test #{@retry_count} before alerting failures." ) + end + + + # + # Write the given message to our logfile - and show it to the console + # if we're running with '--verbose' in play + # + def log_message( msg ) + @logger.info( msg ) + puts msg if ( ENV['VERBOSE'] ) + end + + + + # + # Process jobs from the queue - never return. + # + def run! + while( true ) + log_message( "\n" ) + log_message( "\n" ) + log_message( "Waiting for job.." ) + process_single_job() + end + end + + + + # + # Fetch a single job from the queue, and process it. + # + def process_single_job + + begin + job = @queue.reserve() + + log_message( "Job aquired - Job ID : #{job.id}" ) + + + # + # Parse the JSON of the job body. + # + json = job.body + hash = JSON.parse( json ) + hash['verbose'] = 1 if ( ENV['VERBOSE'] ) + + + # + # Output the details. + # + log_message( "Job body contains the following keys & values:") + hash.keys.each do |key| + log_message( " #{key} => #{hash[key]}" ) + end + + + + # + # Did the test succeed? If not count the number of times it failed in + # a row. We'll repeat several times + # + success = false + count = 0 + + # + # As a result of this test we'll either raise/clear with mauve. + # + # This helper will do that job. + # + alert = Alerter.new( hash ) + + + # + # Convert the test-type to a class name, to do the protocol test. + # + # Given a test-type "foo" we'll attempt to instantiate a class called FOOTest. + # + test = hash['test_type'] + clazz = test.upcase + clazz = "#{clazz}Test" + + + # + # Create the test object. + # + obj = eval(clazz).new( hash ) + + + # + # Ensure that the object we load implements the two methods + # we expect. + # + if ( ( ! obj.respond_to?( "error") ) || + ( ! obj.respond_to?( "run_test" ) ) ) + puts "Class #{clazz} doesn't implement the full protocol-test API" + end + + + + # + # We'll run no more than MAX times. + # + # We stop the execution on a single success. + # + while ( ( count < @retry_count ) && ( success == false ) ) + + log_message( "Running test - attempt #{count}" ) + + if ( obj.run_test() ) + log_message( "Test succeeed - clearing alert" ) + success = true + alert.clear() + end + count += 1 + end + + # + # If we didn't succeed on any of the attempts raise the alert. + # + if ( ! success ) + + # + # Raise the alert, passing the error message. + # + log_message( "Test failed - alerting with #{obj.error()}" ) + alert.raise( obj.error() ) + end + + rescue => ex + puts "Exception raised processing job: #{ex}" + + ensure + # + # Delete the job - either we received an error, in which case + # we should remove it to avoid picking it up again, or we handled + # it successfully so it should be removed. + # + log_message( "Job ID : #{job.id} - Removed" ) + job.delete if ( job ) + end + end +end + + + diff --git a/lib/custodian/alerter.rb b/lib/custodian/alerter.rb new file mode 100644 index 0000000..751b368 --- /dev/null +++ b/lib/custodian/alerter.rb @@ -0,0 +1,76 @@ + + +require 'mauve/sender' +require 'mauve/proto' + + + +# +# This class encapsulates the raising and clearing of alerts via Mauve. +# +class Alerter + + attr_reader :details + + def initialize( test_details ) + @details = test_details + end + + + # + # Raise the alert. + # + def raise( detail ) + + update = Mauve::Proto::AlertUpdate.new + update.alert = [] + update.source = "custodian" + + # be explicit about raising/clearing + update.replace = false + + alert = Mauve::Proto::Alert.new + + # e.g. ping-example.vm.bytemark.co.uk + # e.g. http-http://example.com/page1 + alert.id = "#{@details['test_type']}-#{@details['target_host']}" + + alert.subject = @details['target_host'] + alert.summary = @details['test_alert'] + alert.detail = "The #{@details['test_type']} test failed against #{@details['target_host']}: #{detail}" + alert.raise_time = Time.now.to_i + update.alert << alert + + Mauve::Sender.new("alert.bytemark.co.uk").send(update) + + end + + # + # Clear the alert. + # + def clear + + update = Mauve::Proto::AlertUpdate.new + update.alert = [] + update.source = "custodian" + + # be explicit about raising/clearing + update.replace = false + + + alert = Mauve::Proto::Alert.new + + # e.g. ping-example.vm.bytemark.co.uk + # e.g. http-http://example.com/page1 + alert.id = "#{@details['test_type']}-#{@details['target_host']}" + + alert.subject = @details['target_host'] + alert.summary = @details['test_alert'] + alert.detail = "The #{@details['test_type']} test succeeded against #{@details['target_host']}" + alert.clear_time = Time.now.to_i + update.alert << alert + + Mauve::Sender.new("alert.bytemark.co.uk").send(update) + end + +end -- cgit v1.2.1