god 0.4.3 → 0.5.0

data/History.txt

@@ -1,16 +1,52 @@
-== 0.4.3
+== 0.5.0 / 2007-10-05
+
+* Major Enhancements
+  * Implement lifecycle scoped metric to allow for cross-state conditions
+  * Add TriggerCondition for conditions that need info about state changes
+  * Implement notification system
+  * Add Tasks (a generalization of Watches) to do non-process related tasks
+  * Add example init.d file in GOD_INSTALL_DIR/init/god [scott becker]
+  * Add human readable info to conditions (and make low level log lines debug)
+  * Switch DRb to use a unix domain socket for security reasons
+* Minor Enchancements
+  * Allow EventConditions to do transition overloading
+  * Report errors during god startup instead of failing silently
+  * Make transition block optional (default to Always condition returning true)
+  * Better usage info for `god --help`
+  * Explain what's going on when attempting to rebind to an in-use port
+  * Add -b option to god binary to auto-bind to an unused port
+  * Add `god quit` to stop god without stopping any tasks
+  * Make self-daemonized Watch commands synchronous (as they should be)
+  * Allow self-daemonized Watches to specify a log (could be useful)
+  * Check for existence of config file if specified
+  * Robustify `god load` and report errors back to the command issuer
+  * Warn when `god load` tries to set global options
+  * Add Configurable.clear method and make built-in conditions clear on entry
+* New Conditions
+  * Flapping < TriggerCondition - trigger on state change
+  * HttpResponseCode < PollCondition - trigger on http response code or timeout (thx scott becker)
+* New Contacts
+  * Email < Contact - notify via email (smtp)
+* Bug Fixes
+  * Fix abort not aborting problem
+  * Fix -p option not working for god binary
+  * Fix God.init not accepting block (thx _eric)
+  * Fix SIGHUP ignore (thx _eric)
+  * Fix error reporting on `god --help` (don't error report a normal SystemExit)
+
+== 0.4.3 / 2007-09-10
 * Bug Fixes
-  * fix Process#alive? to not raise on no such file
+  * fix Process#alive? to not raise on no such file (affects `god terminate`)
 
-== 0.4.2
+== 0.4.2 / 2007-09-10
 * Bug Fixes
-  * fix netlink events [dkresge]
+  * fix netlink buffer issue that prevented events on Linux from working consistently [dkresge]
 
-== 0.4.1
+== 0.4.1 / 2007-09-10
 * Bug Fixes
   * require 'stringio' for ruby 1.8.5
 
-== 0.4.0
+== 0.4.0 / 2007-09-10
 
 * Major Enhancements
   * Add the ability for conditions to override transition state (for exceptional cases)
@@ -34,7 +70,7 @@
 * Bug Fixes
   * Use exit!(0) instead of exit! in god binary to exit with code 0 (instead of default -1)
   * Command line group control fixed
-  * Fix cross-thread return problem (use exit instead)  
+  * Fix cross-thread return problem
 
 == 0.3.0 / 2007-08-17
 

data/Manifest.txt

@@ -8,6 +8,7 @@ examples/gravatar.god
 ext/god/extconf.rb
 ext/god/kqueue_handler.c
 ext/god/netlink_handler.c
+init/god
 lib/god.rb
 lib/god/behavior.rb
 lib/god/behaviors/clean_pid_file.rb
@@ -16,11 +17,16 @@ lib/god/condition.rb
 lib/god/conditions/always.rb
 lib/god/conditions/cpu_usage.rb
 lib/god/conditions/degrading_lambda.rb
+lib/god/conditions/flapping.rb
+lib/god/conditions/http_response_code.rb
 lib/god/conditions/lambda.rb
 lib/god/conditions/memory_usage.rb
 lib/god/conditions/process_exits.rb
 lib/god/conditions/process_running.rb
 lib/god/conditions/tries.rb
+lib/god/configurable.rb
+lib/god/contact.rb
+lib/god/contacts/email.rb
 lib/god/dependency_graph.rb
 lib/god/errors.rb
 lib/god/event_handler.rb
@@ -32,19 +38,23 @@ lib/god/logger.rb
 lib/god/metric.rb
 lib/god/process.rb
 lib/god/registry.rb
-lib/god/reporter.rb
-lib/god/server.rb
+lib/god/socket.rb
 lib/god/sugar.rb
 lib/god/system/process.rb
+lib/god/task.rb
 lib/god/timeline.rb
 lib/god/timer.rb
+lib/god/trigger.rb
 lib/god/watch.rb
 test/configs/child_events/child_events.god
 test/configs/child_events/simple_server.rb
 test/configs/child_polls/child_polls.god
 test/configs/child_polls/simple_server.rb
+test/configs/contact/contact.god
+test/configs/contact/simple_server.rb
 test/configs/daemon_events/daemon_events.god
 test/configs/daemon_events/simple_server.rb
+test/configs/daemon_events/simple_server_stop.rb
 test/configs/daemon_polls/daemon_polls.god
 test/configs/daemon_polls/simple_server.rb
 test/configs/degrading_lambda/degrading_lambda.god
@@ -53,12 +63,17 @@ test/configs/real.rb
 test/configs/running_load/running_load.god
 test/configs/stress/simple_server.rb
 test/configs/stress/stress.god
+test/configs/task/logs/.placeholder
+test/configs/task/task.god
 test/configs/test.rb
 test/helper.rb
 test/suite.rb
 test/test_behavior.rb
 test/test_condition.rb
+test/test_conditions_http_response_code.rb
 test/test_conditions_process_running.rb
+test/test_conditions_tries.rb
+test/test_contact.rb
 test/test_dependency_graph.rb
 test/test_event_handler.rb
 test/test_god.rb
@@ -68,10 +83,11 @@ test/test_logger.rb
 test/test_metric.rb
 test/test_process.rb
 test/test_registry.rb
-test/test_reporter.rb
-test/test_server.rb
+test/test_socket.rb
 test/test_sugar.rb
 test/test_system_process.rb
+test/test_task.rb
 test/test_timeline.rb
 test/test_timer.rb
+test/test_trigger.rb
 test/test_watch.rb

data/Rakefile

@@ -1,7 +1,7 @@
 require 'rubygems'
 require 'hoe'
 
-Hoe.new('god', '0.4.3') do |p|
+Hoe.new('god', '0.5.0') do |p|
   p.rubyforge_name = 'god'
   p.author = 'Tom Preston-Werner'
   p.email = 'tom@rubyisawesome.com'

data/bin/god

@@ -1,234 +1,302 @@
 #!/usr/bin/env ruby
 
+STDOUT.sync = true
+
 $:.unshift File.join(File.dirname(__FILE__), *%w[.. lib])
 
 require 'rubygems'
 require 'optparse'
 require 'drb'
 
-options = {:daemonize => true, :port => 17165}
-
-OptionParser.new do |opts|
-  opts.banner = <<-EOF
-Usage: god [command] [options]
-
-  Commands:
-    start <watch or group name>
-    restart <watch or group name>
-    stop <watch or group name>
-    monitor <watch or group name>
-    unmonitor <watch or group name>
-    load <file>
-    log <watch name>
-    status
-    terminate
-
-  Options:
-EOF
-
-  opts.on("-cCONFIG", "--config-file CONFIG", "Configuration file") do |x|
-    options[:config] = x
-  end
-  
-  opts.on("-pPORT", "--port PORT", "Communications port") do |x|
-    options[:port] = x
-  end
-  
-  opts.on("-PFILE", "--pid FILE", "Where to write the PID file") do |x|
-    options[:pid] = x
-  end
-  
-  opts.on("-lFILE", "--log FILE", "Where to write the log file") do |x|
-    options[:log] = x
-  end
-  
-  opts.on("-D", "--no-daemonize", "Don't daemonize") do
-    options[:daemonize] = false
-  end
-  
-  opts.on("-v", "--version", "Print the version number and exit") do
-    options[:version] = true
-  end
-  
-  opts.on("-V", "Print extended version and build information") do
-    options[:info] = true
-  end
-end.parse!
+begin
+  options = {:daemonize => true, :port => 17165}
 
-if options[:version]
-  require 'god'
-  
-  # print version
-  puts "Version #{God::VERSION}"
-  exit!(0)
-elsif options[:info]
-  require 'god'
-  
-  puts "Version: #{God::VERSION}"
-  puts "Polls: enabled"
-  puts "Events: " + God::EventHandler.event_system
-  
-  exit!(0)
-elsif command = ARGV[0]
-  require 'god'
-  
-  # a command was specified
-  
-  # connect to remote drb
-  DRb.start_service
-  server = DRbObject.new nil, "druby://127.0.0.1:#{options[:port]}"
-  
-  begin
-    server.ping
-  rescue DRb::DRbConnError
-    puts "The server is not available (or you do not have permissions to access it)"
-    exit!
-  rescue => e
-    puts e.message
-    puts e.backtrace.join("\n")
-    exit!
-  end
+  opts = OptionParser.new do |opts|
+    opts.banner = <<-EOF
+  Usage:
+    Starting:
+      god -c <config file> [-p <port> | -b] [-P <file>] [-l <file>] [-D]
+      
+    Querying:
+      god <command> <argument> [-p <port>]
+      god <command> [-p <port>]
+      god -v
+      god -V (must be run as root to be accurate on Linux)
+      
+    Commands:
+      start <task or group name>         start task or group
+      restart <task or group name>       restart task or group
+      stop <task or group name>          stop task or group
+      monitor <task or group name>       monitor task or group
+      unmonitor <task or group name>     unmonitor task or group
+      load <file>                        load a config into a running god
+      log <task name>                    show realtime log for given task
+      status                             show status of each task
+      quit                               stop god
+      terminate                          stop god and all tasks
+      
+    Options:
+  EOF
   
-  if command == 'load'
-    file = ARGV[1]
+    opts.on("-cCONFIG", "--config-file CONFIG", "Configuration file") do |x|
+      options[:config] = x
+    end
     
-    puts "Sending '#{command}' command"
+    opts.on("-pPORT", "--port PORT", "Communications port (default 17165)") do |x|
+      options[:port] = x
+    end
     
-    code = File.read(file)
+    opts.on("-b", "--auto-bind", "Auto-bind to an unused port number") do
+      options[:port] = "0"
+    end
     
-    watches = server.running_load(code)
+    opts.on("-PFILE", "--pid FILE", "Where to write the PID file") do |x|
+      options[:pid] = x
+    end
     
-    # output response
-    puts 'The following watches were affected:'
-    watches.each do |w|
-      puts '  ' + w.name
+    opts.on("-lFILE", "--log FILE", "Where to write the log file") do |x|
+      options[:log] = x
     end
     
-    puts "Done"
-  elsif command == 'status'
-    watches = server.status
-    watches.keys.sort.each do |name|
-      state = watches[name][:state]
-      puts "#{name}: #{state}"
+    opts.on("-D", "--no-daemonize", "Don't daemonize") do
+      options[:daemonize] = false
     end
-  elsif command == 'log'
-    begin
-      Signal.trap('INT') { exit!(0) }
-      name = ARGV[1]
-      t = Time.at(0)
-      loop do
-        print server.running_log(name, t)
-        t = Time.now
-        sleep 1
-      end
-    rescue God::NoSuchWatchError
-      puts "No such watch"
-    rescue DRb::DRbConnError
-      puts "The server went away"
-    rescue => e
-      puts e.message
-      puts e.backtrace.join("\n")
-    ensure
-      exit!(0)
+    
+    opts.on("-v", "--version", "Print the version number and exit") do
+      options[:version] = true
     end
-  elsif command == 'terminate'
-    t = Thread.new { loop { STDOUT.print('.'); STDOUT.flush; sleep(1) } }
-    if server.stop_all
-      t.kill; STDOUT.puts
-      puts 'Stopped all watches'
-    else
-      t.kill; STDOUT.puts
-      puts 'Could not stop all watches within 10 seconds'
+    
+    opts.on("-V", "Print extended version and build information") do
+      options[:info] = true
     end
+  end
+  
+  opts.parse!
+  
+  if options[:version]
+    require 'god'
+    
+    # print version
+    puts "Version #{God::VERSION}"
+    exit
+  elsif options[:info]
+    require 'god'
+    
+    puts "Version: #{God::VERSION}"
+    puts "Polls: enabled"
+    puts "Events: " + God::EventHandler.event_system
+    
+    exit
+  elsif command = ARGV[0]
+    require 'god'
+    
+    # a command was specified
+    
+    # connect to drb unix socket
+    DRb.start_service
+    server = DRbObject.new(nil, God::Socket.socket(options[:port]))
     
     begin
-      server.terminate
-      abort 'Could not stop god'
+      server.ping
     rescue DRb::DRbConnError
-      puts 'Stopped god'
-      exit!(0)
+      puts "The server is not available (or you do not have permissions to access it)"
+      abort
     end
-  else
-    # get the name of the watch/group
-    name = ARGV[1]
-  
-    begin
+    
+    if command == 'load'
+      file = ARGV[1]
+      
       puts "Sending '#{command}' command"
+      puts
+      
+      unless File.exist?(file)
+        abort "File not found: #{file}"
+      end
+      
+      names, errors = *server.running_load(File.read(file), File.expand_path(file))
       
-      t = Thread.new { loop { STDOUT.print('.'); STDOUT.flush; sleep(1) } }
-    
-      # send command
-      watches = server.control(name, command)
-    
       # output response
-      t.kill; STDOUT.puts
-      puts 'The following watches were affected:'
-      watches.each do |w|
-        puts '  ' + w.name
+      unless names.empty?
+        puts 'The following tasks were affected:'
+        names.each do |w|
+          puts '  ' + w
+        end
       end
-    rescue God::InvalidCommandError
-      abort "Command '#{command}' is not valid. Run 'god --help' for usage"
-    end
-  end
-  
-  exit!(0)
-else
-  # start god
-  if !options[:daemonize]
-    require 'god'
-    
-    if options[:port]
-      God.port = options[:port]
-    end
-        
-    load File.expand_path(options[:config])
-  else
-    pid = fork do
+      
+      unless errors.empty?
+        puts errors
+        exit(1)
+      end
+    elsif command == 'status'
+      watches = server.status
+      watches.keys.sort.each do |name|
+        state = watches[name][:state]
+        puts "#{name}: #{state}"
+      end
+    elsif command == 'log'
+      begin
+        Signal.trap('INT') { exit }
+        name = ARGV[1]
+        t = Time.at(0)
+        loop do
+          print server.running_log(name, t)
+          t = Time.now
+          sleep 1
+        end
+      rescue God::NoSuchWatchError
+        puts "No such watch"
+      rescue DRb::DRbConnError
+        puts "The server went away"
+      end
+    elsif command == 'quit'
+      begin
+        server.terminate
+        abort 'Could not stop god'
+      rescue DRb::DRbConnError
+        puts 'Stopped god'
+      end
+    elsif command == 'terminate'
+      t = Thread.new { loop { STDOUT.print('.'); STDOUT.flush; sleep(1) } }
+      if server.stop_all
+        t.kill; STDOUT.puts
+        puts 'Stopped all watches'
+      else
+        t.kill; STDOUT.puts
+        puts 'Could not stop all watches within 10 seconds'
+      end
+      
+      begin
+        server.terminate
+        abort 'Could not stop god'
+      rescue DRb::DRbConnError
+        puts 'Stopped god'
+      end
+    else
+      # get the name of the watch/group
+      name = ARGV[1]
+      
       begin
-        require 'god'
+        puts "Sending '#{command}' command"
         
-        log_file = options[:log] || "/dev/null"
+        t = Thread.new { loop { sleep(1); STDOUT.print('.'); STDOUT.flush; sleep(1) } }
         
-        STDIN.reopen "/dev/null"
-        STDOUT.reopen(log_file, "a")
-        STDERR.reopen STDOUT
+        # send command
+        watches = server.control(name, command)
         
-        puts "Starting god"
-      
-        unless God::EventHandler.loaded?
-          puts
-          puts "***********************************************************************"
-          puts "*"
-          puts "* Event conditions are not available for your installation of god."
-          puts "* You may still use and write custom conditions using the poll system"
-          puts "*"
-          puts "***********************************************************************"
-          puts
+        # output response
+        t.kill; STDOUT.puts
+        unless watches.empty?
+          puts 'The following watches were affected:'
+          watches.each do |w|
+            puts '  ' + w
+          end
+        else
+          puts 'No matching task or group'
         end
+      rescue God::InvalidCommandError
+        t.kill rescue nil; STDOUT.puts
+        abort "Command '#{command}' is not valid. Run 'god --help' for usage"
+      end
+    end
+  else
+    # start god
+    $run = true
     
-        puts "Resetting file descriptors"
-        
-        puts "Loading config"
+    if !options[:daemonize]
+      require 'god'
+      
+      if options[:port]
+        God.port = options[:port]
+      end
+      
+      if options[:config]
+        unless File.exist?(options[:config])
+          abort "File not found: #{options[:config]}"
+        end
         
-        if options[:port]
-          God.port = options[:port]
+        begin
+          load File.expand_path(options[:config])
+        rescue Exception => e
+          if e.instance_of?(SystemExit)
+            raise
+          else
+            puts e.message
+            puts e.backtrace.join("\n")
+            abort "There was an error in your configuration file (see above)"
+          end
+        end
+      end
+    else
+      # trap and ignore SIGHUP
+      Signal.trap('HUP') {}
+      
+      pid = fork do
+        begin
+          require 'god'
+          
+          log_file = options[:log] || "/dev/null"
+          
+          unless God::EventHandler.loaded?
+            puts
+            puts "***********************************************************************"
+            puts "*"
+            puts "* Event conditions are not available for your installation of god."
+            puts "* You may still use and write custom conditions using the poll system"
+            puts "*"
+            puts "***********************************************************************"
+            puts
+          end
+          
+          # set port if requested
+          if options[:port]
+            God.port = options[:port]
+          end
+          
+          # load config
+          if options[:config]
+            unless File.exist?(options[:config])
+              abort "File not found: #{options[:config]}"
+            end
+            
+            begin
+              load File.expand_path(options[:config])
+            rescue Exception => e
+              if e.instance_of?(SystemExit)
+                raise
+              else
+                puts e.message
+                puts e.backtrace.join("\n")
+                abort "There was an error in your configuration file (see above)"
+              end
+            end
+          end
+          
+          # reset file descriptors
+          STDIN.reopen "/dev/null"
+          STDOUT.reopen(log_file, "a")
+          STDERR.reopen STDOUT
+        rescue => e
+          puts e.message
+          puts e.backtrace.join("\n")
+          abort "There was a fatal system error while starting god (see above)"
         end
+      end
       
-        load File.expand_path(options[:config])
-        
-        Signal.trap('HUP') {}
-      rescue => e
-        File.open('god.log', 'a') { |f| f.puts e.message + "\n" + e.backtrace }
-        abort "!!! ERROR - See god.log !!!"
+      if options[:pid]
+        File.open(options[:pid], 'w') { |f| f.write pid }
       end
+      
+      ::Process.detach pid
+      
+      exit
     end
-  
-    if options[:pid]
-      File.open(options[:pid], 'w') { |f| f.write pid }
-    end
-  
-    ::Process.detach pid
-  
-    exit!(0)
   end
-end
+rescue Exception => e
+  if e.instance_of?(SystemExit)
+    raise
+  else
+    puts 'Uncaught exception'
+    puts e.message
+    puts e.backtrace.join("\n")
+  end
+end