Allow querying last error that caused lock unhealthiness. Retry test …

…if that last error is retryable.

lib/distributed-lock-google-cloud-storage/lock.rb

@@ -120,6 +120,7 @@ def initialize(bucket_name:, path:, instance_identity_prefix: self.class.default
         @owner = nil
         @metageneration = nil
         @refresher_thread = nil
+        @refresher_error = nil
 
         # The refresher generation is incremented every time we shutdown
         # the refresher thread. It allows the refresher thread to know
@@ -315,6 +316,7 @@ def abandon
       #
       # @return [Boolean]
       # @raise [NotLockedError] This lock was not obtained.
+      # @see #last_refresh_error
       def healthy?
         @state_mutex.synchronize do
           raise NotLockedError, 'Not locked' if !unsynced_locked_according_to_internal_state?
@@ -323,6 +325,8 @@ def healthy?
       end
 
       # Checks whether the lock is healthy. See {#healthy?} for the definition of "healthy".
+      # Use {#last_refresh_error} to query the last error that caused the lock to be declared
+      # unhealthy.
       #
       # It only makes sense to call this method after having obtained this lock.
       #
@@ -333,6 +337,19 @@ def check_health!
         raise LockUnhealthyError, 'Lock is not healthy' if !healthy?
       end
 
+      # Returns the last error that caused the lock to be declared unhealthy.
+      #
+      # Don't use this method to check whether the lock is _currently_ healthy.
+      # If this lock has ever been unhealthy, then this method returns a non-nil value
+      # even if the lock is currently healthy.
+      #
+      # @return [StandardError, nil]
+      def last_refresh_error
+        @state_mutex.synchronize do
+          @refresher_error
+        end
+      end
+
 
       private
 
@@ -565,7 +582,17 @@ def refresh_lock(refresher_generation)
           end
           log_debug { "Done refreshing lock. metageneration=#{file.metageneration}" }
           true
+
         rescue => e
+          @state_mutex.synchronize do
+            if @refresher_generation != refresher_generation
+              log_debug { 'Abort refreshing lock' }
+              return true
+            end
+
+            @refresher_error = e
+          end
+
           log_error { "Error refreshing lock: #{e}" }
           [false, permanent_failure]
         end

spec/lock_spec.rb

@@ -81,6 +81,10 @@ def force_erase_lock_object
       expect(lock).not_to be_owned_according_to_server
     end
 
+    it 'has no refresh error' do
+      expect(lock.last_refresh_error).to be_nil
+    end
+
     specify 'checking for health is not possible due to being unlocked' do
       expect { lock.healthy? }.to \
         raise_error(DistributedLock::GoogleCloudStorage::NotLockedError)
@@ -113,6 +117,7 @@ def force_erase_lock_object
       expect(@lock).to be_owned_according_to_internal_state
       expect(@lock).to be_owned_according_to_server
       expect(@lock).to be_healthy
+      expect(@lock.last_refresh_error).to be_nil
       expect { @lock.check_health! }.not_to raise_error
     end
 
@@ -131,6 +136,7 @@ def force_erase_lock_object
           owned_according_to_internal_state: @lock2.owned_according_to_internal_state?,
           owned_according_to_server: @lock2.owned_according_to_server?,
           healthy: @lock2.healthy?,
+          last_refresh_error: @lock2.last_refresh_error,
         }
       end
 
@@ -155,6 +161,7 @@ def force_erase_lock_object
       expect(result[:owned_according_to_internal_state]).to be_truthy
       expect(result[:owned_according_to_server]).to be_truthy
       expect(result[:healthy]).to be_truthy
+      expect(result[:last_refresh_error]).to be_nil
     end
 
     it 'raises AlreadyLockedError if called twice by the same instance and thread' do
@@ -183,6 +190,7 @@ def force_erase_lock_object
       expect(@lock).to be_owned_according_to_internal_state
       expect(@lock).to be_owned_according_to_server
       expect(@lock).to be_healthy
+      expect(@lock.last_refresh_error).to be_nil
       expect { @lock.check_health! }.not_to raise_error
     end
 
@@ -209,6 +217,7 @@ def force_erase_lock_object
       expect(@lock).to be_owned_according_to_internal_state
       expect(@lock).to be_owned_according_to_server
       expect(@lock).to be_healthy
+      expect(@lock.last_refresh_error).to be_nil
       expect { @lock.check_health! }.not_to raise_error
     end
 
@@ -224,6 +233,7 @@ def force_erase_lock_object
       expect(@lock).to be_owned_according_to_internal_state
       expect(@lock).to be_owned_according_to_server
       expect(@lock).to be_healthy
+      expect(@lock.last_refresh_error).to be_nil
       expect { @lock.check_health! }.not_to raise_error
     end
 
@@ -238,20 +248,25 @@ def force_erase_lock_object
       expect(@lock).to be_owned_according_to_internal_state
       expect(@lock).to be_owned_according_to_server
       expect(@lock).to be_healthy
+      expect(@lock.last_refresh_error).to be_nil
       expect { @lock.check_health! }.not_to raise_error
     end
   end
 
 
   describe '#unlock' do
     after :each do
-      @lock.abandon if @lock
+      if @lock
+        @lock.abandon
+        raise @lock.last_refresh_error if exception_to_retry?(@lock.last_refresh_error)
+      end
     end
 
     def lock_and_unlock
       @lock.lock(timeout: 0)
       deleted = nil
-      expect { deleted = @lock.unlock }.not_to raise_error
+      e = catch_non_retryable_exception { deleted = @lock.unlock }
+      expect(e).to be_nil
       deleted
     end
 
@@ -292,8 +307,11 @@ def lock_and_unlock
 
       @lock.lock(timeout: 0)
       force_erase_lock_object
+
       deleted = nil
-      expect { deleted = @lock.unlock }.not_to raise_error
+      e = catch_non_retryable_exception { deleted = @lock.unlock }
+
+      expect(e).to be_nil
       expect(deleted).to be_falsey
       expect(@lock).not_to be_locked_according_to_internal_state
       expect(@lock).not_to be_locked_according_to_server
@@ -326,7 +344,10 @@ def lock_and_unlock
     end
 
     after :each do
-      @lock.abandon if @lock
+      if @lock
+        @lock.abandon
+        raise @lock.last_refresh_error if exception_to_retry?(@lock.last_refresh_error)
+      end
     end
 
     it 'updates the update time' do
@@ -348,12 +369,13 @@ def lock_and_unlock
         f.metadata['something'] = '123'
       end
       expect(file.metageneration).not_to eq(orig_metageneration)
-      eventually(timeout: 10) do
+      eventually(timeout: 30) do
         !@lock.healthy?
       end
       expect { @lock.check_health! }.to \
         raise_error(DistributedLock::GoogleCloudStorage::LockUnhealthyError)
       expect(log_output.string).to include('Lock object has an unexpected metageneration number')
+      expect(@lock.last_refresh_error).not_to be_nil
     end
 
     it 'declares unhealthiness when the lock object is deleted' do
@@ -370,6 +392,7 @@ def lock_and_unlock
       expect { @lock.check_health! }.to \
         raise_error(DistributedLock::GoogleCloudStorage::LockUnhealthyError)
       expect(log_output.string).to include('Lock object has been unexpectedly deleted')
+      expect(@lock.last_refresh_error).not_to be_nil
     end
   end
 end

spec/spec_helper.rb

@@ -5,6 +5,11 @@
 require 'rspec/retry'
 
 module Helpers
+  EXCEPTIONS_TO_RETRY = [
+    Google::Cloud::ResourceExhaustedError,
+    Google::Cloud::UnavailableError,
+  ]
+
   def self.print_all_thread_backtraces
     puts '-------- Begin backtrace dump --------'
     Thread.list.each do |thr|
@@ -17,6 +22,33 @@ def self.print_all_thread_backtraces
     puts '-------- End backtrace dump --------'
   end
 
+  def exception_to_retry?(ex)
+    EXCEPTIONS_TO_RETRY.any? do |ex_class|
+      ex.is_a?(ex_class)
+    end
+  end
+
+  # Runs the given block.
+  # If it raises a retryable exception, then pass it through.
+  # If it raises a non-retryable exception, then returns that exception.
+  # Otherwise, returns nil.
+  #
+  # This is used as an alternative to `expect { block }.not_to raise_error`
+  # which lets retryable exceptions through so that rspec-retry can retry
+  # the test.
+  def catch_non_retryable_exception
+    begin
+      yield
+      nil
+    rescue => e
+      if exception_to_retry?(e)
+        raise e
+      else
+        e
+      end
+    end
+  end
+
   def require_envvar(name)
     value = ENV[name]
     raise ArgumentError, "Required environment variable: #{name}" if value.to_s.empty?
@@ -56,10 +88,7 @@ def consistently(duration:, interval: 0.1)
   c.display_try_failure_messages = true
   c.default_retry_count = 3
   c.default_sleep_interval = 10
-  c.exceptions_to_retry = [
-    Google::Cloud::ResourceExhaustedError,
-    Google::Cloud::UnavailableError
-  ]
+  c.exceptions_to_retry = Helpers::EXCEPTIONS_TO_RETRY
 end
 
 Signal.trap('QUIT') { Helpers.print_all_thread_backtraces }