Fix arv-run-pipeline-instance to not crash when template is null. no issue #
[arvados.git] / sdk / cli / bin / arv-run-pipeline-instance
index 980ce83ae3a42c153d0a445d01fac4a3cf0c4dd4..c6fd61f9e3c90fe4836b9918635db9bdde984a1e 100755 (executable)
@@ -42,6 +42,8 @@
 # [--status-json path] Print JSON status report to a file or
 #                      fifo. Default: /dev/null
 #
+# [--description] Description for the pipeline instance.
+#
 # == Parameters
 #
 # [param_name=param_value]
@@ -175,6 +177,10 @@ p = Trollop::Parser.new do
       "Manage the pipeline in process.",
       :short => :none,
       :type => :boolean)
+  opt(:description,
+      "Description for the pipeline instance.",
+      :short => :none,
+      :type => :string)
   stop_on [:'--']
 end
 $options = Trollop::with_standard_exception_handling p do
@@ -478,6 +484,8 @@ class WhRunPipelineInstance
         end
       end
     else
+      description = $options[:description]
+      description = ("Created at #{Time.now.localtime}" + (@template[:name].andand.size>0 ? " using the pipeline template *#{@template[:name]}*" : "")) if !description
       @instance = PipelineInstance.
         create(components: @components,
                properties: {
@@ -486,6 +494,7 @@ class WhRunPipelineInstance
                  }
                },
                pipeline_template_uuid: @template[:uuid],
+               description: description,
                state: ($options[:submit] ? 'RunningOnServer' : 'RunningOnClient'))
     end
     self
@@ -495,6 +504,10 @@ class WhRunPipelineInstance
     moretodo = true
     interrupted = false
 
+    if @instance[:started_at].nil?
+      @instance[:started_at] = Time.now
+    end
+
     job_creation_failed = 0
     while moretodo
       moretodo = false
@@ -518,7 +531,6 @@ class WhRunPipelineInstance
             :script_version => c[:script_version],
             :repository => c[:repository],
             :nondeterministic => c[:nondeterministic],
-            :output_is_persistent => c[:output_is_persistent] || false,
             :runtime_constraints => c[:runtime_constraints],
             :owner_uuid => owner_uuid,
           }, {
@@ -562,37 +574,54 @@ class WhRunPipelineInstance
               # This is my first time discovering that the job
               # succeeded. (At the top of this loop, I was still
               # waiting for it to finish.)
-              if c[:output_is_persistent]
-                # I need to make sure a resources/wants link is in
-                # place to protect the output from garbage
-                # collection. (Normally Crunch does this for me, but
-                # here I might be reusing the output of someone else's
-                # job and I need to make sure it's understood that the
-                # output is valuable to me, too.)
-                wanted = c[:job][:output]
-                debuglog "checking for existing persistence link for #{wanted}"
-                @my_user_uuid ||= $arv.user.current[:uuid]
-                links = $arv.link.list(limit: 1,
-                                       filters:
-                                       [%w(link_class = resources),
-                                        %w(name = wants),
-                                        %w(tail_uuid =) + [@my_user_uuid],
-                                        %w(head_uuid =) + [wanted]
-                                       ])[:items]
-                if links.any?
-                  debuglog "link already exists, uuid #{links.first[:uuid]}"
+
+              debuglog "names: #{@instance[:name]} #{@template[:name]}", 0
+              if (not @instance[:name].nil?) and (not @instance[:name].empty?)
+                pipeline_name = @instance[:name]
+              elsif @instance[:pipeline_template_uuid]
+                fetch_template(@instance[:pipeline_template_uuid])
+                pipeline_name = @template[:name]
+              else
+                pipeline_name = "pipeline started #{@instance[:started_at]}"
+              end
+              if c[:output_name] != false
+                # Create a collection located in the same project as the pipeline with the contents of the output.
+                portable_data_hash = c[:job][:output]
+                collections = $arv.collection.list(limit: 1,
+                                                   filters: [['portable_data_hash', '=', portable_data_hash]],
+                                                   select: ["portable_data_hash", "manifest_text"]
+                                                   )[:items]
+                if collections.any?
+                  name = c[:output_name] || "Output #{portable_data_hash[0..7]} of #{cname} of #{pipeline_name}"
+
+                  # check if there is a name collision.
+                  name_collisions = $arv.collection.list(filters: [["owner_uuid", "=", owner_uuid],
+                                                                   ["name", "=", name]])[:items]
+
+                  newcollection_actual = nil
+                  if name_collisions.any? and name_collisions.first[:portable_data_hash] == portable_data_hash
+                    # There is already a collection with the same name and the
+                    # same contents, so just point to that.
+                    newcollection_actual = name_collisions.first
+                  end
+
+                  if newcollection_actual.nil?
+                    # Did not find a collection with the same name (or the
+                    # collection has a different portable data hash) so create
+                    # a new collection with ensure_unique_name: true.
+                    newcollection = {
+                      owner_uuid: owner_uuid,
+                      name: name,
+                      portable_data_hash: collections.first[:portable_data_hash],
+                      manifest_text: collections.first[:manifest_text]
+                    }
+                    debuglog "Creating collection #{newcollection}", 0
+                    newcollection_actual = $arv.collection.create collection: newcollection, ensure_unique_name: true
+                  end
+
+                  c[:output_uuid] = newcollection_actual[:uuid]
                 else
-                  newlink = $arv.link.create link: \
-                  {
-                    link_class: 'resources',
-                    name: 'wants',
-                    tail_kind: 'arvados#user',
-                    tail_uuid: @my_user_uuid,
-                    head_kind: 'arvados#collection',
-                    head_uuid: wanted,
-                    owner_uuid: owner_uuid
-                  }
-                  debuglog "added link, uuid #{newlink[:uuid]}"
+                  debuglog "Could not find a collection with portable data hash #{portable_data_hash}", 0
                 end
               end
             end
@@ -665,6 +694,10 @@ class WhRunPipelineInstance
       end
     end
 
+    if @instance[:finished_at].nil? and ['Complete', 'Failed'].include? @instance[:state]
+      @instance[:finished_at] = Time.now
+    end
+
     debuglog "pipeline instance state is #{@instance[:state]}"
 
     # set components_summary
@@ -728,6 +761,7 @@ class WhRunPipelineInstance
       if ["New", "Ready", "RunningOnClient",
           "RunningOnServer"].include?(@instance[:state])
         @instance[:state] = "Failed"
+        @instance[:finished_at] = Time.now
         @instance.save
       end
       @instance.log_stderr(msg)