bricolage-streamingload 0.9.0 → 0.10.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8bddfd0337158c01e5acb0fac9ff9d4b3e152029
-  data.tar.gz: 7243c54d80c5bcd4c8e157fd8d5a14e66e629abe
+  metadata.gz: 4a46109dbac16f28b807d4e2b29808d254ebe209
+  data.tar.gz: f97bf6c58057d05f89f6fe36b1b23b18b5b1d089
 SHA512:
-  metadata.gz: 3e85208c4fcbf1a199169e75b40490181127fbaf7398e6ebb78daaead3f36704e76b7177ce43edfe404e5509e95f5a592ed8db4d4f7fa379efcca0d382fdedd6
-  data.tar.gz: eb8187f0b900731bdf21c4e3b3ffbb0ac24ad133bd6d7819b20e4b7eb24420e649bf690d43f0b0e3aa5caf496146e208a0d13fa33c6be72183b8469eca96bb1c
+  metadata.gz: 4bc1033f8c06cb6f56a782129d1927919f3689828710182494a03edccc750b9dd4ab915ba7efc7f79255e754d36f4c466ebd6189fce9de1dee6985b1b6cfecce
+  data.tar.gz: aef6bd65662f942641ea2701880c41e28c88caa8dcb9aab9e02c93a2d76d28fdd0aec28dc0c4d4b159a3c2098520da8cfad27f9d0cd1824ae40428c3f5b91988

data/lib/bricolage/streamingload/job.rb

@@ -18,9 +18,10 @@ module Bricolage
 
     class Job
 
-      def initialize(context:, ctl_ds:, log_table: 'strload_load_logs', task_id:, force: false, logger:)
+      def initialize(context:, ctl_ds:, data_ds:, log_table: 'strload_load_logs', task_id:, force: false, logger:)
         @context = context
         @ctl_ds = ctl_ds
+        @data_ds = data_ds
         @log_table = log_table
         @task_id = task_id
         @force = force
@@ -28,7 +29,6 @@ module Bricolage
 
         @task = nil
         @job_id = nil
-        @data_ds = nil
         @manifest = nil
       end
 
@@ -57,9 +57,7 @@ module Bricolage
       rescue DataConnectionFailed => ex
         @logger.error ex.message
         wait_for_connection('data', @data_ds) unless fail_fast
-        # FIXME: tmp: We don't know the transaction was succeeded or not in the Redshift, auto-retry is too dangerous.
-        #return false
-        return true
+        return false
       rescue JobFailure => ex
         @logger.error ex.message
         return false
@@ -87,6 +85,16 @@ module Bricolage
             raise JobDefered, "defered: task_id=#{@task_id}"
           end
 
+          if @task.unknown_state?
+            true_status = DataConnection.open(@data_ds, @logger) {|data|
+              data.get_job_status(@log_table, @task.last_job_id)
+            }
+            @logger.info "fixiating unknown job status: job_id=#{@task.last_job_id}, status=(unknown->#{true_status})"
+            @task.fix_last_job_status true_status
+            ctl.fix_job_status @task.last_job_id, true_status
+            @logger.info "job status fixed."
+          end
+
           @job_id = ctl.begin_job(@task_id, @process_id, @force)
           unless @job_id
             @logger.warn "task is already succeeded and not forced; discard task: task_id=#{@task_id}"
@@ -102,17 +110,14 @@ module Bricolage
           }
         rescue ControlConnectionFailed
           raise
-
-        # FIXME: tmp: should be a failure, not an error.
         rescue DataConnectionFailed => ex
           ctl.open {
-            ctl.abort_job job_id, 'error', ex.message.lines.first.strip
+            ctl.abort_job job_id, 'unknown', ex.message.lines.first.strip
           }
           raise
-
         rescue JobFailure => ex
           ctl.open {
-            fail_count = ctl.fail_count(@task_id)
+            fail_count = @task.failure_count
             final_retry = (fail_count >= MAX_RETRY)
             retry_msg = (fail_count > 0) ? "(retry\##{fail_count}#{final_retry ? ' FINAL' : ''}) " : ''
             ctl.abort_job job_id, 'failure', retry_msg + ex.message.lines.first.strip
@@ -193,6 +198,11 @@ module Bricolage
           raise DataConnectionFailed, "data connection failed: #{ex.message}"
         end
 
+        def get_job_status(log_table, job_id)
+          count = @connection.query_value("select count(*) from #{log_table} where job_id = #{job_id}")
+          count.to_i > 0 ? 'success' : 'failure'
+        end
+
         def load_with_work_table(work_table, manifest, options, sql_source, log_table, job_id)
           @connection.transaction {|txn|
             # NOTE: This transaction ends with truncation, this DELETE does nothing
@@ -255,7 +265,32 @@ module Bricolage
           raise ControlConnectionFailed, "control connection failed: #{ex.message}"
         end
 
-        TaskInfo = Struct.new(:task_id, :task_class, :schema_name, :table_name, :disabled, :object_urls)
+        TaskInfo = Struct.new(:task_id, :task_class, :schema_name, :table_name, :disabled, :object_urls, :jobs)
+        class TaskInfo
+          def unknown_state?
+            return false if jobs.empty?
+            jobs.last.status == 'unknown'
+          end
+
+          def last_job_id
+            return nil if jobs.empty?
+            jobs.last.job_id
+          end
+
+          def fix_last_job_status(st)
+            jobs.last.status = st unless jobs.empty?
+          end
+
+          def failure_count
+            @failure_count ||= begin
+              statuses = jobs.map(&:status)
+              statuses.delete('duplicated')
+              last_succ = statuses.rindex('success')
+              statuses[0..last_succ] = [] if last_succ
+              statuses.size
+            end
+          end
+        end
 
         def load_task(task_id)
           rec = @connection.query_row(<<-EndSQL) or raise JobError, "no such task: #{task_id}"
@@ -277,10 +312,29 @@ module Bricolage
             rec['schema_name'],
             rec['table_name'],
             (rec['disabled'] != 'f'),
-            load_object_urls(task_id)
+            load_object_urls(task_id),
+            load_jobs(task_id)
           )
         end
 
+        def load_jobs(task_id)
+          records = @connection.query_rows(<<-EndSQL)
+            select
+                job_id
+                , status
+            from
+                strload_jobs
+            where
+                task_id = #{task_id}
+            order by
+                start_time
+            ;
+          EndSQL
+          records.map {|rec| JobInfo.new(rec['job_id'].to_i, rec['status']) }
+        end
+
+        JobInfo = Struct.new(:job_id, :status, :start_time)
+
         def load_object_urls(task_id)
           urls = @connection.query_values(<<-EndSQL)
             select
@@ -296,6 +350,20 @@ module Bricolage
           urls
         end
 
+        def fix_job_status(job_id, status)
+          @connection.update(<<-EndSQL)
+            update
+                strload_jobs
+            set
+                status = #{s status}
+                , message = 'status fixed: ' || message
+            where
+                job_id = #{job_id}
+                and status = 'unknown'
+            ;
+          EndSQL
+        end
+
         def begin_job(task_id, process_id, force)
           job_id = @connection.query_value(<<-EndSQL)
             insert into strload_jobs
@@ -320,22 +388,6 @@ module Bricolage
           return job_id ? job_id.to_i : nil
         end
 
-        def fail_count(task_id)
-          statuses = @connection.query_values(<<-EndSQL)
-            select
-                j.status
-            from
-                strload_tasks t
-                inner join strload_jobs j using (task_id)
-            where
-                t.task_id = #{task_id}
-            order by
-                j.job_id desc
-          EndSQL
-          statuses.shift if statuses.first == 'running'   # current job
-          statuses.take_while {|st| %w[failure error].include?(st) }.size
-        end
-
         def commit_job(job_id, message = nil)
           @connection.transaction {|txn|
             write_job_result job_id, 'success', (message || '')

data/lib/bricolage/streamingload/taskhandler.rb

@@ -29,6 +29,7 @@ module Bricolage
         ctx = Context.for_application(opts.working_dir, environment: opts.environment, logger: logger)
 
         ctl_ds = ctx.get_data_source('sql', config.fetch('ctl-postgres-ds', 'db_ctl'))
+        data_ds = ctx.get_data_source('sql', config.fetch('redshift-ds', 'db_data'))
         task_queue = ctx.get_data_source('sqs', config.fetch('task-queue-ds', 'sqs_task'))
         log_table = config.fetch('log-table', 'strload_load_logs')
         service_logger =
@@ -41,6 +42,7 @@ module Bricolage
         task_handler = new(
           context: ctx,
           ctl_ds: ctl_ds,
+          data_ds: data_ds,
           log_table: log_table,
           task_queue: task_queue,
           working_dir: opts.working_dir,
@@ -92,9 +94,10 @@ module Bricolage
         # ignore
       end
 
-      def initialize(context:, ctl_ds:, log_table:, task_queue:, working_dir:, logger:, job_class: Job)
+      def initialize(context:, ctl_ds:, data_ds:, log_table:, task_queue:, working_dir:, logger:, job_class: Job)
         @ctx = context
         @ctl_ds = ctl_ds
+        @data_ds = data_ds
         @log_table = log_table
         @task_queue = task_queue
         @working_dir = working_dir
@@ -132,7 +135,7 @@ module Bricolage
       end
 
       def new_job(task_id, force)
-        @job_class.new(context: @ctx, ctl_ds: @ctl_ds, log_table: @log_table, task_id: task_id, force: force, logger: @logger)
+        @job_class.new(context: @ctx, ctl_ds: @ctl_ds, data_ds: data_ds, log_table: @log_table, task_id: task_id, force: force, logger: @logger)
       end
 
       def job_class

data/lib/bricolage/streamingload/version.rb

@@ -1,5 +1,5 @@
 module Bricolage
   module StreamingLoad
-    VERSION = '0.9.0'
+    VERSION = '0.10.0'
   end
 end

data/test/streamingload/test_job.rb

@@ -16,7 +16,7 @@ module Bricolage
     class TestJob < Test::Unit::TestCase
 
       test "execute_task" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.desttable', 'testschema', 'desttable', 100, 1800, false]
           db.insert_into 'strload_tasks', [1, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [1, 1], [1, 2]
@@ -24,7 +24,7 @@ module Bricolage
             [1, 's3://data-bucket/testschema.desttable/0001.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp],
             [2, 's3://data-bucket/testschema.desttable/0002.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 1, force: false, logger: ctx.logger)
+          job = new_job(task_id: 1, force: false)
           job.execute_task
 
           assert_equal [
@@ -42,7 +42,7 @@ module Bricolage
       end
 
       test "execute_task (with work table)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.with_work_table', 'testschema', 'with_work_table', 100, 1800, false]
           db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
@@ -50,7 +50,7 @@ module Bricolage
             [1001, 's3://data-bucket/testschema.with_work_table/0001.json.gz', 1024, 'testschema.with_work_table', 'mmmm', current_timestamp, current_timestamp],
             [1002, 's3://data-bucket/testschema.with_work_table/0002.json.gz', 1024, 'testschema.with_work_table', 'mmmm', current_timestamp, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: false, logger: ctx.logger)
+          job = new_job(task_id: 11, force: false)
           job.execute_task
 
           assert_equal [
@@ -70,11 +70,11 @@ module Bricolage
       end
 
       test "execute_task (disabled)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.desttable', 'testschema', 'desttable', 100, 1800, true]
           db.insert_into 'strload_tasks', [1, 'streaming_load_v3', 1, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 1, force: false, logger: ctx.logger)
+          job = new_job(task_id: 1, force: false)
           assert_raise(JobDefered) {
             job.execute_task
           }
@@ -84,7 +84,7 @@ module Bricolage
       end
 
       test "execute_task (duplicated)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.desttable', 'testschema', 'desttable', 100, 1800, false]
           db.insert_into 'strload_tasks', [1, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_jobs',
@@ -92,7 +92,7 @@ module Bricolage
             [2, 1, 'localhost-1234', 'success', current_timestamp, current_timestamp, ''],
             [3, 1, 'localhost-1234', 'duplicated', current_timestamp, current_timestamp, '']
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 1, force: false, logger: ctx.logger)
+          job = new_job(task_id: 1, force: false)
           assert_raise(JobDuplicated) {
             job.execute_task
           }
@@ -100,7 +100,7 @@ module Bricolage
       end
 
       test "execute_task (duplicated but forced)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.desttable', 'testschema', 'desttable', 100, 1800, false]
           db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
@@ -108,7 +108,7 @@ module Bricolage
             [1001, 's3://data-bucket/testschema.desttable/0001.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp],
             [1002, 's3://data-bucket/testschema.desttable/0002.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: true, logger: ctx.logger)
+          job = new_job(task_id: 11, force: true)
           job.execute_task
 
           assert_equal [
@@ -127,7 +127,7 @@ module Bricolage
       end
 
       test "execute_task (load fails / first time)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.sql_fails', 'testschema', 'sql_fails', 100, 1800, false]
           db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
@@ -135,7 +135,7 @@ module Bricolage
             [1001, 's3://data-bucket/testschema.desttable/0001.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp],
             [1002, 's3://data-bucket/testschema.desttable/0002.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: false, logger: ctx.logger)
+          job = new_job(task_id: 11, force: false)
           assert_raise(JobFailure) {
             job.execute_task
           }
@@ -153,7 +153,7 @@ module Bricolage
       end
 
       test "execute_task (load fails / nth time)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.sql_fails', 'testschema', 'sql_fails', 100, 1800, false]
           db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
@@ -164,7 +164,7 @@ module Bricolage
             [101, 11, 'localhost-1234', 'failure', current_timestamp, current_timestamp, 'query failed'],
             [102, 11, 'localhost-1234', 'failure', current_timestamp, current_timestamp, 'query failed']
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: false, logger: ctx.logger)
+          job = new_job(task_id: 11, force: false)
           assert_raise(JobFailure) {
             job.execute_task
           }
@@ -183,7 +183,7 @@ module Bricolage
       end
 
       test "execute_task (too many retry)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.sql_fails', 'testschema', 'sql_fails', 100, 1800, false]
           db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
@@ -197,7 +197,7 @@ module Bricolage
             [104, 11, 'localhost-1234', 'failure', current_timestamp, current_timestamp, 'retry#3 query failed'],
             [105, 11, 'localhost-1234', 'failure', current_timestamp, current_timestamp, 'retry#4 query failed']
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: false, logger: ctx.logger)
+          job = new_job(task_id: 11, force: false)
           assert_raise(JobCancelled) {
             job.execute_task
           }
@@ -216,7 +216,7 @@ module Bricolage
       end
 
       test "execute_task (job error)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.job_error', 'testschema', 'job_error', 100, 1800, false]
           db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
@@ -224,7 +224,7 @@ module Bricolage
             [1001, 's3://data-bucket/testschema.job_error/0001.json.gz', 1024, 'testschema.job_error', 'mmmm', current_timestamp, current_timestamp],
             [1002, 's3://data-bucket/testschema.job_error/0002.json.gz', 1024, 'testschema.job_error', 'mmmm', current_timestamp, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: false, logger: ctx.logger)
+          job = new_job(task_id: 11, force: false)
           assert_raise(JobError) {
             job.execute_task
           }
@@ -242,15 +242,15 @@ module Bricolage
       end
 
       test "execute_task (unexpected error)" do
-        setup_context {|ctx, ctl_ds, db|
+        setup_context {|db|
           db.insert_into 'strload_tables', [1, 'testschema.unexpected_error', 'testschema', 'unexpected_error', 100, 1800, false]
-          db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, sql('current_timestamp')]
+          db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
           db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
           db.insert_into 'strload_objects',
             [1001, 's3://data-bucket/testschema.unexpected_error/0001.json.gz', 1024, 'testschema.unexpected_error', 'mmmm', current_timestamp, current_timestamp],
             [1002, 's3://data-bucket/testschema.unexpected_error/0002.json.gz', 1024, 'testschema.unexpected_error', 'mmmm', current_timestamp, current_timestamp]
 
-          job = Job.new(context: ctx, ctl_ds: ctl_ds, task_id: 11, force: false, logger: ctx.logger)
+          job = new_job(task_id: 11, force: false)
           assert_raise(JobError) {
             job.execute_task
           }
@@ -267,16 +267,78 @@ module Bricolage
         }
       end
 
+      test "execute_task (unknown status, really=success)" do
+        setup_context {|db|
+          db.insert_into 'strload_tables', [1, 'testschema.desttable', 'testschema', 'desttable', 100, 1800, false]
+          db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
+          db.insert_into 'strload_jobs',
+            [101, 11, 'localhost-1234', 'unknown', current_timestamp, current_timestamp, 'data connection failed']
+          @data_ds.provide_job_status 101, true
+
+          job = new_job(task_id: 11, force: false)
+          assert_raise(JobDuplicated) {
+            job.execute_task
+          }
+
+          job_row = db.query_row("select * from strload_jobs where job_id = 101")
+          assert_equal 'success', job_row['status']
+        }
+      end
+
+      test "execute_task (unknown status, really=failure)" do
+        setup_context {|db|
+          db.insert_into 'strload_tables', [1, 'testschema.desttable', 'testschema', 'desttable', 100, 1800, false]
+          db.insert_into 'strload_tasks', [11, 'streaming_load_v3', 1, current_timestamp]
+          db.insert_into 'strload_task_objects', [11, 1001], [11, 1002]
+          db.insert_into 'strload_objects',
+            [1001, 's3://data-bucket/testschema.desttable/0001.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp],
+            [1002, 's3://data-bucket/testschema.desttable/0002.json.gz', 1024, 'testschema.desttable', 'mmmm', current_timestamp, current_timestamp]
+          db.insert_into 'strload_jobs',
+            [101, 11, 'localhost-1234', 'unknown', current_timestamp, current_timestamp, 'data connection failed']
+          @data_ds.provide_job_status 101, false
+
+          job = new_job(task_id: 11, force: false)
+          job.execute_task
+
+          assert_equal [
+            "begin transaction;",
+            "copy testschema.desttable from '#{job.manifest.url}' credentials 'cccc' manifest statupdate false compupdate false json 'auto' gzip timeformat 'auto' dateformat 'auto' acceptanydate acceptinvchars ' ' truncatecolumns trimblanks ;",
+            "insert into strload_load_logs (job_id, finish_time) values (#{job.job_id}, current_timestamp)",
+            "commit;"
+          ], job.data_ds.sql_list
+
+          job_row = db.query_row("select * from strload_jobs where job_id = 101")
+          assert_equal 'failure', job_row['status']
+
+          job_row = db.query_row("select * from strload_jobs where job_id = #{job.job_id}")
+          assert_equal 11, job_row['task_id'].to_i
+          assert_equal job.process_id, job_row['process_id']
+          assert_equal 'success', job_row['status']
+        }
+      end
+
       def setup_context(verbose: false)
-        ctx = Context.for_application('.', environment: 'test', logger: (verbose ? nil : NullLogger.new))
-        ctl_ds = ctx.get_data_source('sql', 'dwhctl')
-        ctl_ds.open {|conn|
+        @ctx = Context.for_application('.', environment: 'test', logger: (verbose ? nil : NullLogger.new))
+        @ctl_ds = @ctx.get_data_source('sql', 'dwhctl')
+        @data_ds = @ctx.get_data_source('sql', 'db_data_mock')
+        @ctl_ds.open {|conn|
           client = SQLClient.new(conn)
           clear_all_tables(client)
-          yield ctx, ctl_ds, client
+          yield client
         }
       end
 
+      def new_job(task_id:, force:)
+        Job.new(
+          context: @ctx,
+          ctl_ds: @ctl_ds,
+          data_ds: @data_ds,
+          logger: @ctx.logger,
+          task_id: task_id,
+          force: force
+        )
+      end
+
       # FIXME: database cleaner
       def clear_all_tables(client)
         client.truncate_tables %w[
@@ -373,6 +435,7 @@ module Bricolage
           @fail_pattern = fail_pattern ? Regexp.compile(fail_pattern) : nil
           @error_pattern = error_pattern ? Regexp.compile(error_pattern) : nil
           @exception_pattern = exception_pattern ? Regexp.compile(exception_pattern) : nil
+          @job_status = {}
         end
 
         attr_reader :sql_list
@@ -400,11 +463,30 @@ module Bricolage
           end
         end
 
+        def provide_job_status(job_id, succeeded)
+          @job_status[job_id] = succeeded
+        end
+
+        def job_succeeded?(job_id)
+          raise "job status unregistered: job_id=#{job_id}" unless @job_status.key?(job_id)
+          @job_status[job_id]
+        end
+
         class Connection
           def initialize(ds)
             @ds = ds
           end
 
+          def query_value(sql)
+            case sql
+            when /\bstrload_load_logs where job_id = (\d+)/
+              job_id = $1.to_i
+              @ds.job_succeeded?(job_id) ? 1 : 0
+            else
+              raise "unknown query: #{sql}"
+            end
+          end
+
           def execute(sql)
             @ds.issue_sql sql
           end
@@ -482,7 +564,27 @@ module Bricolage
         end
       end
 
-    end
+      test "TaskInfo#failure_count" do
+        test_data = [
+          [%w[], 0],
+          [%w[success], 0],
+          [%w[failure], 1],
+          [%w[error], 1],
+          [%w[failure failure], 2],
+          [%w[failure error], 2],
+          [%w[failure success], 0],
+          [%w[success success], 0],
+          [%w[failure success failure], 1],
+          [%w[failure success failure success failure failure], 2]
+        ]
+        c = Job::ControlConnection
+        test_data.each do |status_list, expected_count|
+          task = c::TaskInfo.new(nil,nil,nil,nil,nil,nil, status_list.map {|st| c::JobInfo.new(nil, st) })
+          assert_equal expected_count, task.failure_count
+        end
+      end
+
+    end   # class TestJob
 
-  end
-end
+  end   # module StreamingLoad
+end   # module Bricolage

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bricolage-streamingload
 version: !ruby/object:Gem::Version
-  version: 0.9.0
+  version: 0.10.0
 platform: ruby
 authors:
 - Minero Aoki