From: radhika <radhika@curoverse.com>
Date: Tue, 13 Dec 2016 19:05:15 +0000 (-0500)
Subject: 10028: upgrade full text search indexes to not include leading space.
X-Git-Tag: 1.1.0~533^2
X-Git-Url: https://git.arvados.org/arvados.git/commitdiff_plain/d13cc16ebbe287fe81d06c5faa1d15fe588bcf16?ds=sidebyside

10028: upgrade full text search indexes to not include leading space.
---

diff --git a/services/api/app/models/arvados_model.rb b/services/api/app/models/arvados_model.rb
index aed0309591..910db7e0c7 100644
--- a/services/api/app/models/arvados_model.rb
+++ b/services/api/app/models/arvados_model.rb
@@ -252,12 +252,7 @@ class ArvadosModel < ActiveRecord::Base
     parts = full_text_searchable_columns.collect do |column|
       "coalesce(#{column},'')"
     end
-    # We prepend a space to the tsvector() argument here. Otherwise,
-    # it might start with a column that has its own (non-full-text)
-    # index, which causes Postgres to use the column index instead of
-    # the tsvector index, which causes full text queries to be just as
-    # slow as if we had no index at all.
-    "to_tsvector('english', ' ' || #{parts.join(" || ' ' || ")})"
+    "to_tsvector('english', #{parts.join(" || ' ' || ")})"
   end
 
   def self.apply_filters query, filters
diff --git a/services/api/app/models/job.rb b/services/api/app/models/job.rb
index ef3d0b5e10..9556799aba 100644
--- a/services/api/app/models/job.rb
+++ b/services/api/app/models/job.rb
@@ -119,6 +119,10 @@ class Job < ArvadosModel
     super - ["script_parameters_digest"]
   end
 
+  def self.full_text_searchable_columns
+    super - ["script_parameters_digest"]
+  end
+
   def self.load_job_specific_filters attrs, orig_filters, read_users
     # Convert Job-specific @filters entries into general SQL filters.
     script_info = {"repository" => nil, "script" => nil}
diff --git a/services/api/db/migrate/20161213172944_full_text_search_indexes.rb b/services/api/db/migrate/20161213172944_full_text_search_indexes.rb
new file mode 100644
index 0000000000..4cb4d978a1
--- /dev/null
+++ b/services/api/db/migrate/20161213172944_full_text_search_indexes.rb
@@ -0,0 +1,32 @@
+class FullTextSearchIndexes < ActiveRecord::Migration
+  def fts_indexes
+    {
+      "collections" => "collections_full_text_search_idx",
+      "container_requests" => "container_requests_full_text_search_idx",
+      "groups" => "groups_full_text_search_idx",
+      "jobs" => "jobs_full_text_search_idx",
+      "pipeline_instances" => "pipeline_instances_full_text_search_idx",
+      "pipeline_templates" => "pipeline_templates_full_text_search_idx",
+      "workflows" => "workflows_full_text_search_idx",
+    }
+  end
+
+  def up
+    # remove existing fts indexes and create up to date ones with no leading space
+    fts_indexes.each do |t, i|
+      ActiveRecord::Base.connection.indexes(t).each do |idx|
+        if idx.name == i
+          remove_index t.to_sym, :name => i
+          break
+        end
+      end
+      execute "CREATE INDEX #{i} ON #{t} USING gin(#{t.classify.constantize.full_text_tsvector});"
+    end
+  end
+
+  def down
+    fts_indexes.each do |t, i|
+      remove_index t.to_sym, :name => i
+    end
+  end
+end
diff --git a/services/api/db/structure.sql b/services/api/db/structure.sql
index e715cd60c4..7ee7ea6ff9 100644
--- a/services/api/db/structure.sql
+++ b/services/api/db/structure.sql
@@ -1506,7 +1506,7 @@ CREATE UNIQUE INDEX collection_owner_uuid_name_unique ON collections USING btree
 -- Name: collections_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX collections_full_text_search_idx ON collections USING gin (to_tsvector('english'::regconfig, (((((((((((((((((' '::text || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(portable_data_hash, ''::character varying))::text) || ' '::text) || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || (COALESCE(description, ''::character varying))::text) || ' '::text) || COALESCE(properties, ''::text)) || ' '::text) || (COALESCE(file_names, ''::character varying))::text)));
+CREATE INDEX collections_full_text_search_idx ON collections USING gin (to_tsvector('english'::regconfig, (((((((((((((((((COALESCE(owner_uuid, ''::character varying))::text || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(portable_data_hash, ''::character varying))::text) || ' '::text) || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || (COALESCE(description, ''::character varying))::text) || ' '::text) || COALESCE(properties, ''::text)) || ' '::text) || (COALESCE(file_names, ''::character varying))::text)));
 
 
 --
@@ -1520,7 +1520,7 @@ CREATE INDEX collections_search_index ON collections USING btree (owner_uuid, mo
 -- Name: container_requests_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX container_requests_full_text_search_idx ON container_requests USING gin (to_tsvector('english'::regconfig, (((((((((((((((((((((((((((((((((((' '::text || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(description, ''::text)) || ' '::text) || COALESCE(properties, ''::text)) || ' '::text) || (COALESCE(state, ''::character varying))::text) || ' '::text) || (COALESCE(requesting_container_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(container_uuid, ''::character varying))::text) || ' '::text) || COALESCE(mounts, ''::text)) || ' '::text) || COALESCE(runtime_constraints, ''::text)) || ' '::text) || (COALESCE(container_image, ''::character varying))::text) || ' '::text) || COALESCE(environment, ''::text)) || ' '::text) || (COALESCE(cwd, ''::character varying))::text) || ' '::text) || COALESCE(command, ''::text)) || ' '::text) || (COALESCE(output_path, ''::character varying))::text) || ' '::text) || COALESCE(filters, ''::text))));
+CREATE INDEX container_requests_full_text_search_idx ON container_requests USING gin (to_tsvector('english'::regconfig, (((((((((((((((((((((((((((((((((((((((((COALESCE(uuid, ''::character varying))::text || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(description, ''::text)) || ' '::text) || COALESCE(properties, ''::text)) || ' '::text) || (COALESCE(state, ''::character varying))::text) || ' '::text) || (COALESCE(requesting_container_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(container_uuid, ''::character varying))::text) || ' '::text) || COALESCE(mounts, ''::text)) || ' '::text) || COALESCE(runtime_constraints, ''::text)) || ' '::text) || (COALESCE(container_image, ''::character varying))::text) || ' '::text) || COALESCE(environment, ''::text)) || ' '::text) || (COALESCE(cwd, ''::character varying))::text) || ' '::text) || COALESCE(command, ''::text)) || ' '::text) || (COALESCE(output_path, ''::character varying))::text) || ' '::text) || COALESCE(filters, ''::text)) || ' '::text) || COALESCE(scheduling_parameters, ''::text)) || ' '::text) || (COALESCE(output_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(log_uuid, ''::character varying))::text)));
 
 
 --
@@ -1541,7 +1541,7 @@ CREATE INDEX containers_search_index ON containers USING btree (uuid, owner_uuid
 -- Name: groups_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX groups_full_text_search_idx ON groups USING gin (to_tsvector('english'::regconfig, (((((((((((((' '::text || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || (COALESCE(description, ''::character varying))::text) || ' '::text) || (COALESCE(group_class, ''::character varying))::text)));
+CREATE INDEX groups_full_text_search_idx ON groups USING gin (to_tsvector('english'::regconfig, (((((((((((((COALESCE(uuid, ''::character varying))::text || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || (COALESCE(description, ''::character varying))::text) || ' '::text) || (COALESCE(group_class, ''::character varying))::text)));
 
 
 --
@@ -2283,7 +2283,7 @@ CREATE INDEX job_tasks_search_index ON job_tasks USING btree (uuid, owner_uuid,
 -- Name: jobs_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX jobs_full_text_search_idx ON jobs USING gin (to_tsvector('english'::regconfig, (((((((((((((((((((((((((((((((((((((((((' '::text || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(submit_id, ''::character varying))::text) || ' '::text) || (COALESCE(script, ''::character varying))::text) || ' '::text) || (COALESCE(script_version, ''::character varying))::text) || ' '::text) || COALESCE(script_parameters, ''::text)) || ' '::text) || (COALESCE(cancelled_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(cancelled_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(output, ''::character varying))::text) || ' '::text) || (COALESCE(is_locked_by_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(log, ''::character varying))::text) || ' '::text) || COALESCE(tasks_summary, ''::text)) || ' '::text) || COALESCE(runtime_constraints, ''::text)) || ' '::text) || (COALESCE(repository, ''::character varying))::text) || ' '::text) || (COALESCE(supplied_script_version, ''::character varying))::text) || ' '::text) || (COALESCE(docker_image_locator, ''::character varying))::text) || ' '::text) || (COALESCE(description, ''::character varying))::text) || ' '::text) || (COALESCE(state, ''::character varying))::text) || ' '::text) || (COALESCE(arvados_sdk_version, ''::character varying))::text)));
+CREATE INDEX jobs_full_text_search_idx ON jobs USING gin (to_tsvector('english'::regconfig, (((((((((((((((((((((((((((((((((((((((((((COALESCE(uuid, ''::character varying))::text || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(submit_id, ''::character varying))::text) || ' '::text) || (COALESCE(script, ''::character varying))::text) || ' '::text) || (COALESCE(script_version, ''::character varying))::text) || ' '::text) || COALESCE(script_parameters, ''::text)) || ' '::text) || (COALESCE(cancelled_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(cancelled_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(output, ''::character varying))::text) || ' '::text) || (COALESCE(is_locked_by_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(log, ''::character varying))::text) || ' '::text) || COALESCE(tasks_summary, ''::text)) || ' '::text) || COALESCE(runtime_constraints, ''::text)) || ' '::text) || (COALESCE(repository, ''::character varying))::text) || ' '::text) || (COALESCE(supplied_script_version, ''::character varying))::text) || ' '::text) || (COALESCE(docker_image_locator, ''::character varying))::text) || ' '::text) || (COALESCE(description, ''::character varying))::text) || ' '::text) || (COALESCE(state, ''::character varying))::text) || ' '::text) || (COALESCE(arvados_sdk_version, ''::character varying))::text) || ' '::text) || COALESCE(components, ''::text))));
 
 
 --
@@ -2339,7 +2339,7 @@ CREATE INDEX nodes_search_index ON nodes USING btree (uuid, owner_uuid, modified
 -- Name: pipeline_instances_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX pipeline_instances_full_text_search_idx ON pipeline_instances USING gin (to_tsvector('english'::regconfig, (((((((((((((((((((((' '::text || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(pipeline_template_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(components, ''::text)) || ' '::text) || COALESCE(properties, ''::text)) || ' '::text) || (COALESCE(state, ''::character varying))::text) || ' '::text) || COALESCE(components_summary, ''::text)) || ' '::text) || (COALESCE(description, ''::character varying))::text)));
+CREATE INDEX pipeline_instances_full_text_search_idx ON pipeline_instances USING gin (to_tsvector('english'::regconfig, (((((((((((((((((((((COALESCE(uuid, ''::character varying))::text || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(pipeline_template_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(components, ''::text)) || ' '::text) || COALESCE(properties, ''::text)) || ' '::text) || (COALESCE(state, ''::character varying))::text) || ' '::text) || COALESCE(components_summary, ''::text)) || ' '::text) || (COALESCE(description, ''::character varying))::text)));
 
 
 --
@@ -2360,7 +2360,7 @@ CREATE UNIQUE INDEX pipeline_template_owner_uuid_name_unique ON pipeline_templat
 -- Name: pipeline_templates_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX pipeline_templates_full_text_search_idx ON pipeline_templates USING gin (to_tsvector('english'::regconfig, (((((((((((((' '::text || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(components, ''::text)) || ' '::text) || (COALESCE(description, ''::character varying))::text)));
+CREATE INDEX pipeline_templates_full_text_search_idx ON pipeline_templates USING gin (to_tsvector('english'::regconfig, (((((((((((((COALESCE(uuid, ''::character varying))::text || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(components, ''::text)) || ' '::text) || (COALESCE(description, ''::character varying))::text)));
 
 
 --
@@ -2416,7 +2416,7 @@ CREATE INDEX virtual_machines_search_index ON virtual_machines USING btree (uuid
 -- Name: workflows_full_text_search_idx; Type: INDEX; Schema: public; Owner: -; Tablespace: 
 --
 
-CREATE INDEX workflows_full_text_search_idx ON workflows USING gin (to_tsvector('english'::regconfig, (((((((((((((' '::text || (COALESCE(uuid, ''::character varying))::text) || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(description, ''::text)) || ' '::text) || COALESCE(definition, ''::text))));
+CREATE INDEX workflows_full_text_search_idx ON workflows USING gin (to_tsvector('english'::regconfig, (((((((((((((COALESCE(uuid, ''::character varying))::text || ' '::text) || (COALESCE(owner_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_client_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(modified_by_user_uuid, ''::character varying))::text) || ' '::text) || (COALESCE(name, ''::character varying))::text) || ' '::text) || COALESCE(description, ''::text)) || ' '::text) || COALESCE(definition, ''::text))));
 
 
 --
@@ -2704,4 +2704,6 @@ INSERT INTO schema_migrations (version) VALUES ('20161111143147');
 
 INSERT INTO schema_migrations (version) VALUES ('20161115171221');
 
-INSERT INTO schema_migrations (version) VALUES ('20161115174218');
\ No newline at end of file
+INSERT INTO schema_migrations (version) VALUES ('20161115174218');
+
+INSERT INTO schema_migrations (version) VALUES ('20161213172944');
\ No newline at end of file