Add an option to raise on failure

libraries/choregraphie.rb

@@ -129,7 +129,7 @@ def self.ensure_whyrun_supported(run_context, resource_name, ignore_missing_reso
       begin
         resource = run_context.resource_collection.find(resource_name)
       rescue Chef::Exceptions::ResourceNotFound
-        if ignore_missing_resource # rubocop:disable Style/GuardClause
+        if ignore_missing_resource
           # some resources are defined only when used
           # so we ignore them
           return

libraries/primitive_consul_lock.rb

@@ -18,6 +18,7 @@ def initialize(options = {})
       raise ArgumentError, "You can't set both concurrency and service" if @options[:concurrency] && @options[:service]
 
       @options[:backoff] ||= 5 # seconds
+      @options[:raise_on_failures] = false
 
       ConsulCommon.setup_consul(@options)
 
@@ -74,7 +75,9 @@ def wait_until(action, opts = {})
       dc = "(in #{@options[:datacenter]})" if @options[:datacenter]
       Chef::Log.info "Will #{action} the lock #{path} #{dc}"
       start = Time.now
-      success = 0.upto(opts[:max_failures] || Float::INFINITY).any? do |tries|
+      max_failures = opts[:max_failures] || @options[:max_failures]
+      raise_on_failures = opts[:raise_on_failures].nil? ? @options[:raise_on_failures] : opts[:raise_on_failures]
+      success = 0.upto(max_failures || Float::INFINITY).any? do |tries|
         yield(start, tries) || backoff(start, tries)
       rescue StandardError => e
         Chef::Log.warn "Error while #{action}-ing lock"
@@ -85,7 +88,8 @@ def wait_until(action, opts = {})
       if success
         Chef::Log.info "#{action.to_s.capitalize}ed the lock #{path}"
       else
-        Chef::Log.warn "Will ignore errors and since we've reached #{opts[:max_failures]} errors"
+        Chef::Log.warn "Will ignore errors and since we've reached #{max_failures} errors"
+        raise "Max attempt #{max_failures} reached" if max_failures && raise_on_failures
       end
     end
 
@@ -101,7 +105,7 @@ def register(choregraphie)
         # The reason we have to be a bit more relaxed here, is that all
         # chef run including a choregraphie with this primitive try to
         # release the lock at the end of a successful run
-        wait_until(:exit, max_failures: 5) { semaphore.exit(name: @options[:id]) }
+        wait_until(:exit, max_failures: 5, raise_on_failures: false) { semaphore.exit(name: @options[:id]) }
       end
     end
 

libraries/primitive_consul_maintenance.rb

@@ -28,7 +28,7 @@ def maintenance?
       # we observed in very rare cases that maintenance status could be not up to date
       # for ~5s after a restart
       results = 3.times.map do
-        checks = Diplomat::Agent.checks()
+        checks = Diplomat::Agent.checks
         maint_status = checks.dig(@maintenance_key, 'Status')
         sleep(@options[:check_interval])
         maint_status == 'critical' && !checks.dig(@maintenance_key, 'Notes').nil?
@@ -40,7 +40,7 @@ def maintenance?
 
     # @return [String, nil] reason of the maintenance, if any. Nil otherwise
     def maintenance_reason
-      checks = Diplomat::Agent.checks()
+      checks = Diplomat::Agent.checks
       checks.dig(@maintenance_key, 'Notes')
     end
 

libraries/primitive_consul_rack_lock.rb

@@ -27,7 +27,7 @@ def register(choregraphie)
         # The reason we have to be a bit more relaxed here, is that all
         # chef run including a choregraphie with this primitive try to
         # release the lock at the end of a successful run
-        wait_until(:exit, max_failures: 5) { semaphore.exit(name: @options[:rack], server: @options[:id]) }
+        wait_until(:exit, max_failures: 5, raise_on_failures: false) { semaphore.exit(name: @options[:rack], server: @options[:id]) }
       end
     end