pghero 1.5.3 → 2.0.0

data/lib/pghero/methods/query_stats.rb

@@ -1,41 +1,42 @@
 module PgHero
   module Methods
     module QueryStats
-      def query_stats(options = {})
-        current_query_stats = options[:historical] && options[:end_at] && options[:end_at] < Time.now ? [] : current_query_stats(options)
-        historical_query_stats = options[:historical] ? historical_query_stats(options) : []
+      def query_stats(historical: false, start_at: nil, end_at: nil, min_average_time: nil, min_calls: nil, **options)
+        current_query_stats = historical && end_at && end_at < Time.now ? [] : current_query_stats(options)
+        historical_query_stats = historical && historical_query_stats_enabled? ? historical_query_stats(start_at: start_at, end_at: end_at, **options) : []
 
-        query_stats = combine_query_stats((current_query_stats + historical_query_stats).group_by { |q| [q["query_hash"], q["user"]] })
-        query_stats = combine_query_stats(query_stats.group_by { |q| [normalize_query(q["query"]), q["user"]] })
+        query_stats = combine_query_stats((current_query_stats + historical_query_stats).group_by { |q| [q[:query_hash], q[:user]] })
+        query_stats = combine_query_stats(query_stats.group_by { |q| [normalize_query(q[:query]), q[:user]] })
 
         # add percentages
-        all_queries_total_minutes = [current_query_stats, historical_query_stats].sum { |s| (s.first || {})["all_queries_total_minutes"].to_f }
+        all_queries_total_minutes = [current_query_stats, historical_query_stats].sum { |s| (s.first || {})[:all_queries_total_minutes] || 0 }
         query_stats.each do |query|
-          query["average_time"] = query["total_minutes"] * 1000 * 60 / query["calls"]
-          query["total_percent"] = query["total_minutes"] * 100.0 / all_queries_total_minutes
+          query[:average_time] = query[:total_minutes] * 1000 * 60 / query[:calls]
+          query[:total_percent] = query[:total_minutes] * 100.0 / all_queries_total_minutes
         end
 
         sort = options[:sort] || "total_minutes"
-        query_stats = query_stats.sort_by { |q| -q[sort] }.first(100)
-        if options[:min_average_time]
-          query_stats.reject! { |q| q["average_time"].to_f < options[:min_average_time] }
+        query_stats = query_stats.sort_by { |q| -q[sort.to_sym] }.first(100)
+        if min_average_time
+          query_stats.reject! { |q| q[:average_time] < min_average_time }
         end
-        if options[:min_calls]
-          query_stats.reject! { |q| q["calls"].to_i < options[:min_calls] }
+        if min_calls
+          query_stats.reject! { |q| q[:calls] < min_calls }
         end
         query_stats
       end
 
       def query_stats_available?
-        select_all("SELECT COUNT(*) AS count FROM pg_available_extensions WHERE name = 'pg_stat_statements'").first["count"].to_i > 0
+        select_one("SELECT COUNT(*) AS count FROM pg_available_extensions WHERE name = 'pg_stat_statements'") > 0
       end
 
+      # only cache if true
       def query_stats_enabled?
-        query_stats_extension_enabled? && query_stats_readable?
+        @query_stats_enabled ||= query_stats_readable?
       end
 
       def query_stats_extension_enabled?
-        select_all("SELECT COUNT(*) AS count FROM pg_extension WHERE extname = 'pg_stat_statements'").first["count"].to_i > 0
+        select_one("SELECT COUNT(*) AS count FROM pg_extension WHERE extname = 'pg_stat_statements'") > 0
       end
 
       def query_stats_readable?
@@ -46,7 +47,8 @@ module PgHero
       end
 
       def enable_query_stats
-        execute("CREATE EXTENSION pg_stat_statements")
+        execute("CREATE EXTENSION IF NOT EXISTS pg_stat_statements")
+        true
       end
 
       def disable_query_stats
@@ -55,46 +57,29 @@ module PgHero
       end
 
       def reset_query_stats
-        if query_stats_enabled?
-          execute("SELECT pg_stat_statements_reset()")
-          true
-        else
-          false
-        end
+        execute("SELECT pg_stat_statements_reset()")
+        true
+      rescue ActiveRecord::StatementInvalid
+        false
       end
 
       # http://stackoverflow.com/questions/20582500/how-to-check-if-a-table-exists-in-a-given-schema
       def historical_query_stats_enabled?
         # TODO use schema from config
-        PgHero.truthy?(stats_connection.select_all(squish <<-SQL
-          SELECT EXISTS (
-            SELECT
-              1
-            FROM
-              pg_catalog.pg_class c
-            INNER JOIN
-              pg_catalog.pg_namespace n ON n.oid = c.relnamespace
-            WHERE
-              n.nspname = 'public'
-              AND c.relname = 'pghero_query_stats'
-              AND c.relkind = 'r'
-          )
-        SQL
-        ).to_a.first["exists"]) && capture_query_stats?
+        # make sure primary database is PostgreSQL first
+        query_stats_table_exists? && capture_query_stats? && !missing_query_stats_columns.any?
       end
 
-      def supports_query_hash?
-        @supports_query_hash ||= server_version_num >= 90400 && historical_query_stats_enabled? && PgHero::QueryStats.column_names.include?("query_hash")
+      def query_stats_table_exists?
+        table_exists?("pghero_query_stats")
       end
 
-      def supports_query_stats_user?
-        @supports_query_stats_user ||= historical_query_stats_enabled? && PgHero::QueryStats.column_names.include?("user")
+      def missing_query_stats_columns
+        %w(query_hash user) - PgHero::QueryStats.column_names
       end
 
-      def insert_stats(table, columns, values)
-        values = values.map { |v| "(#{v.map { |v2| quote(v2) }.join(",")})" }.join(",")
-        columns = columns.map { |v| quote_table_name(v) }.join(",")
-        stats_connection.execute("INSERT INTO #{quote_table_name(table)} (#{columns}) VALUES #{values}")
+      def supports_query_hash?
+        server_version_num >= 90400
       end
 
       # resetting query stats will reset across the entire Postgres instance
@@ -121,51 +106,68 @@ module PgHero
         if query_stats.any? { |_, v| v.any? } && reset_query_stats
           query_stats.each do |db_id, db_query_stats|
             if db_query_stats.any?
-              supports_query_hash = PgHero.databases[db_id].supports_query_hash?
-              supports_query_stats_user = PgHero.databases[db_id].supports_query_stats_user?
-
               values =
                 db_query_stats.map do |qs|
-                  values = [
+                  [
                     db_id,
-                    qs["query"],
-                    qs["total_minutes"].to_f * 60 * 1000,
-                    qs["calls"],
-                    now
+                    qs[:query],
+                    qs[:total_minutes] * 60 * 1000,
+                    qs[:calls],
+                    now,
+                    qs[:query_hash],
+                    qs[:user]
                   ]
-                  values << qs["query_hash"] if supports_query_hash
-                  values << qs["user"] if supports_query_stats_user
-                  values
                 end
 
-              columns = %w[database query total_time calls captured_at]
-              columns << "query_hash" if supports_query_hash
-              columns << "user" if supports_query_stats_user
-
+              columns = %w[database query total_time calls captured_at query_hash user]
               insert_stats("pghero_query_stats", columns, values)
             end
           end
         end
       end
 
-      private
+      def slow_queries(query_stats: nil, **options)
+        query_stats ||= self.query_stats(options)
+        query_stats.select { |q| q[:calls].to_i >= slow_query_calls.to_i && q[:average_time].to_f >= slow_query_ms.to_f }
+      end
 
-      def stats_connection
-        ::PgHero::QueryStats.connection
+      def query_hash_stats(query_hash)
+        if historical_query_stats_enabled? && supports_query_hash?
+          start_at = 24.hours.ago
+          select_all_stats <<-SQL
+            SELECT
+              captured_at,
+              total_time / 1000 / 60 AS total_minutes,
+              (total_time / calls) AS average_time,
+              calls,
+              (SELECT regexp_matches(query, '/\\*(.+)\\*/'))[1] AS origin
+            FROM
+              pghero_query_stats
+            WHERE
+              database = #{quote(id)}
+              AND captured_at >= #{quote(start_at)}
+              AND query_hash = #{quote(query_hash)}
+            ORDER BY
+              1 ASC
+          SQL
+        else
+          raise NotEnabled, "Query hash stats not enabled"
+        end
       end
 
+      private
+
       # http://www.craigkerstiens.com/2013/01/10/more-on-postgres-performance/
-      def current_query_stats(options = {})
+      def current_query_stats(limit: nil, sort: nil, database: nil, query_hash: nil)
         if query_stats_enabled?
-          limit = options[:limit] || 100
-          sort = options[:sort] || "total_minutes"
-          database = options[:database] ? quote(options[:database]) : "current_database()"
+          limit ||= 100
+          sort ||= "total_minutes"
           select_all <<-SQL
             WITH query_stats AS (
               SELECT
                 LEFT(query, 10000) AS query,
                 #{supports_query_hash? ? "queryid" : "md5(query)"} AS query_hash,
-                #{supports_query_stats_user? ? "rolname" : "NULL"} AS user,
+                rolname AS user,
                 (total_time / 1000 / 60) AS total_minutes,
                 (total_time / calls) AS average_time,
                 calls
@@ -176,7 +178,8 @@ module PgHero
               INNER JOIN
                 pg_roles ON pg_roles.oid = pg_stat_statements.userid
               WHERE
-                pg_database.datname = #{database}
+                pg_database.datname = #{database ? quote(database) : "current_database()"}
+                #{query_hash ? "AND queryid = #{quote(query_hash)}" : nil}
             )
             SELECT
               query,
@@ -194,19 +197,19 @@ module PgHero
             LIMIT #{limit.to_i}
           SQL
         else
-          []
+          raise NotEnabled, "Query stats not enabled"
         end
       end
 
-      def historical_query_stats(options = {})
+      def historical_query_stats(sort: nil, start_at: nil, end_at: nil, query_hash: nil)
         if historical_query_stats_enabled?
-          sort = options[:sort] || "total_minutes"
-          stats_connection.select_all squish <<-SQL
+          sort ||= "total_minutes"
+          select_all_stats <<-SQL
             WITH query_stats AS (
               SELECT
                 #{supports_query_hash? ? "query_hash" : "md5(query)"} AS query_hash,
-                #{supports_query_stats_user? ? "pghero_query_stats.user" : "NULL"} AS user,
-                array_agg(LEFT(query, 10000)) AS query,
+                pghero_query_stats.user AS user,
+                array_agg(LEFT(query, 10000) ORDER BY REPLACE(LEFT(query, 1000), '?', '!') COLLATE "C" ASC) AS query,
                 (SUM(total_time) / 1000 / 60) AS total_minutes,
                 (SUM(total_time) / SUM(calls)) AS average_time,
                 SUM(calls) AS calls
@@ -215,15 +218,17 @@ module PgHero
               WHERE
                 database = #{quote(id)}
                 #{supports_query_hash? ? "AND query_hash IS NOT NULL" : ""}
-                #{options[:start_at] ? "AND captured_at >= #{quote(options[:start_at])}" : ""}
-                #{options[:end_at] ? "AND captured_at <= #{quote(options[:end_at])}" : ""}
+                #{start_at ? "AND captured_at >= #{quote(start_at)}" : ""}
+                #{end_at ? "AND captured_at <= #{quote(end_at)}" : ""}
+                #{query_hash ? "AND query_hash = #{quote(query_hash)}" : ""}
               GROUP BY
                 1, 2
             )
             SELECT
               query_hash,
               query_stats.user,
-              query[1],
+              query[1] AS query,
+              query[array_length(query, 1)] AS explainable_query,
               total_minutes,
               average_time,
               calls,
@@ -236,31 +241,32 @@ module PgHero
             LIMIT 100
           SQL
         else
-          []
+          raise NotEnabled, "Historical query stats not enabled"
         end
       end
 
-      def server_version_num
-        @server_version ||= select_all("SHOW server_version_num").first["server_version_num"].to_i
-      end
-
       def combine_query_stats(grouped_stats)
         query_stats = []
         grouped_stats.each do |_, stats2|
           value = {
-            "query" => (stats2.find { |s| s["query"] } || {})["query"],
-            "user" => (stats2.find { |s| s["user"] } || {})["user"],
-            "query_hash" => (stats2.find { |s| s["query"] } || {})["query_hash"],
-            "total_minutes" => stats2.sum { |s| s["total_minutes"].to_f },
-            "calls" => stats2.sum { |s| s["calls"].to_i },
-            "all_queries_total_minutes" => stats2.sum { |s| s["all_queries_total_minutes"].to_f }
+            query: (stats2.find { |s| s[:query] } || {})[:query],
+            user: (stats2.find { |s| s[:user] } || {})[:user],
+            query_hash: (stats2.find { |s| s[:query_hash] } || {})[:query_hash],
+            total_minutes: stats2.sum { |s| s[:total_minutes] },
+            calls: stats2.sum { |s| s[:calls] }.to_i,
+            all_queries_total_minutes: stats2.sum { |s| s[:all_queries_total_minutes] }
           }
-          value["total_percent"] = value["total_minutes"] * 100.0 / value["all_queries_total_minutes"]
+          value[:total_percent] = value[:total_minutes] * 100.0 / value[:all_queries_total_minutes]
+          value[:explainable_query] = stats2.map { |s| s[:explainable_query] }.select { |q| q && explainable?(q) }.first
           query_stats << value
         end
         query_stats
       end
 
+      def explainable?(query)
+        query =~ /select/i && !query.include?("?)") && !query.include?("= ?") && !query.include?("$1") && query !~ /limit \?/i
+      end
+
       # removes comments
       # combines ?, ?, ? => ?
       def normalize_query(query)

data/lib/pghero/methods/{replica.rb → replication.rb}

@@ -1,13 +1,16 @@
 module PgHero
   module Methods
-    module Replica
+    module Replication
       def replica?
-        select_all("SELECT setting FROM pg_settings WHERE name = 'hot_standby'").first["setting"] == "on"
+        unless defined?(@replica)
+          @replica = select_one("SELECT pg_is_in_recovery()")
+        end
+        @replica
       end
 
       # http://www.postgresql.org/message-id/CADKbJJWz9M0swPT3oqe8f9+tfD4-F54uE6Xtkh4nERpVsQnjnw@mail.gmail.com
       def replication_lag
-        select_all(<<-SQL
+        select_one <<-SQL
           SELECT
             CASE
               WHEN pg_last_xlog_receive_location() = pg_last_xlog_replay_location() THEN 0
@@ -15,7 +18,20 @@ module PgHero
             END
           AS replication_lag
         SQL
-        ).first["replication_lag"].to_f
+      end
+
+      def replication_slots
+        select_all <<-SQL
+          SELECT
+            slot_name,
+            database,
+            active
+          FROM pg_replication_slots
+        SQL
+      end
+
+      def replicating?
+        select_all("SELECT state FROM pg_stat_replication").any?
       end
     end
   end

data/lib/pghero/methods/sequences.rb

@@ -21,16 +21,15 @@ module PgHero
             sequence_name ASC
         SQL
 
-        select_all(sequences.map { |s| "SELECT last_value FROM #{s["sequence"]}" }.join(" UNION ALL ")).each_with_index do |row, i|
-          sequences[i]["last_value"] = row["last_value"]
+        select_all(sequences.map { |s| "SELECT last_value FROM #{s[:sequence]}" }.join(" UNION ALL ")).each_with_index do |row, i|
+          sequences[i][:last_value] = row[:last_value]
         end
 
         sequences
       end
 
-      def sequence_danger(options = {})
-        threshold = (options[:threshold] || 0.9).to_f
-        sequences.select { |s| s["last_value"].to_i / s["max_value"].to_f > threshold }.sort_by { |s| s["max_value"].to_i - s["last_value"].to_i }
+      def sequence_danger(threshold: 0.9)
+        sequences.select { |s| s[:last_value] / s[:max_value].to_f > threshold }.sort_by { |s| s[:max_value] - s[:last_value] }
       end
     end
   end

data/lib/pghero/methods/space.rb

@@ -2,40 +2,133 @@ module PgHero
   module Methods
     module Space
       def database_size
-        select_all("SELECT pg_size_pretty(pg_database_size(current_database()))").first["pg_size_pretty"]
+        PgHero.pretty_size select_one("SELECT pg_database_size(current_database())")
       end
 
       def relation_sizes
-        select_all <<-SQL
+        select_all_size <<-SQL
           SELECT
             n.nspname AS schema,
-            c.relname AS name,
+            c.relname AS relation,
             CASE WHEN c.relkind = 'r' THEN 'table' ELSE 'index' END AS type,
-            pg_size_pretty(pg_table_size(c.oid)) AS size,
             pg_table_size(c.oid) AS size_bytes
           FROM
             pg_class c
           LEFT JOIN
-            pg_namespace n ON (n.oid = c.relnamespace)
+            pg_namespace n ON n.oid = c.relnamespace
           WHERE
             n.nspname NOT IN ('pg_catalog', 'information_schema')
             AND n.nspname !~ '^pg_toast'
             AND c.relkind IN ('r', 'i')
           ORDER BY
             pg_table_size(c.oid) DESC,
-            name ASC
+            2 ASC
         SQL
       end
 
+      def table_sizes
+        select_all_size <<-SQL
+          SELECT
+            n.nspname AS schema,
+            c.relname AS table,
+            pg_total_relation_size(c.oid) AS size_bytes
+          FROM
+            pg_class c
+          LEFT JOIN
+            pg_namespace n ON n.oid = c.relnamespace
+          WHERE
+            n.nspname NOT IN ('pg_catalog', 'information_schema')
+            AND n.nspname !~ '^pg_toast'
+            AND c.relkind = 'r'
+          ORDER BY
+            pg_total_relation_size(c.oid) DESC,
+            2 ASC
+        SQL
+      end
+
+      def space_growth(days: 7, relation_sizes: nil)
+        if space_stats_enabled?
+          relation_sizes ||= self.relation_sizes
+          sizes = Hash[ relation_sizes.map { |r| [r[:relation], r[:size_bytes]] } ]
+          start_at = days.days.ago
+
+          stats = select_all_stats <<-SQL
+            WITH t AS (
+              SELECT
+                relation,
+                array_agg(size ORDER BY captured_at) AS sizes
+              FROM
+                pghero_space_stats
+              WHERE
+                database = #{quote(id)}
+                AND captured_at >= #{quote(start_at)}
+              GROUP BY
+                1
+            )
+            SELECT
+              relation,
+              sizes[1] AS size_bytes
+            FROM
+              t
+            ORDER BY
+              1
+          SQL
+
+          stats.each do |r|
+            relation = r[:relation]
+            if sizes[relation]
+              r[:growth_bytes] = sizes[relation] - r[:size_bytes]
+            end
+            r.delete(:size_bytes)
+          end
+          stats
+        else
+          raise NotEnabled, "Space stats not enabled"
+        end
+      end
+
+      def relation_space_stats(relation)
+        if space_stats_enabled?
+          relation_sizes ||= self.relation_sizes
+          sizes = Hash[ relation_sizes.map { |r| [r[:relation], r[:size_bytes]] } ]
+          start_at = 30.days.ago
+
+          stats = select_all_stats <<-SQL
+            SELECT
+              captured_at,
+              size AS size_bytes
+            FROM
+              pghero_space_stats
+            WHERE
+              database = #{quote(id)}
+              AND captured_at >= #{quote(start_at)}
+              AND relation = #{quote(relation)}
+            ORDER BY
+              1 ASC
+          SQL
+
+          stats << {
+            captured_at: Time.now,
+            size_bytes: sizes[relation].to_i
+          }
+        else
+          raise NotEnabled, "Space stats not enabled"
+        end
+      end
+
       def capture_space_stats
         now = Time.now
-        columns = %w[database schema relation size captured_at]
+        columns = %w(database schema relation size captured_at)
         values = []
         relation_sizes.each do |rs|
-          values << [id, rs["schema"], rs["name"], rs["size_bytes"].to_i, now]
+          values << [id, rs[:schema], rs[:name], rs[:size_bytes].to_i, now]
         end
         insert_stats("pghero_space_stats", columns, values)
       end
+
+      def space_stats_enabled?
+        table_exists?("pghero_space_stats")
+      end
     end
   end
 end