feat: add pg13 + pg_partman docker

README.md

@@ -4,8 +4,12 @@
 
 Extract matomo data from [`Live.getLastVisitsDetails`](https://developer.matomo.org/api-reference/reporting-api) API and push events and visits informations to Postgres.
 
+Use [pg_partman](https://github.com/pgpartman/pg_partman) to partition data by month.
+
 ## Usage
 
+Create the [initial table](./initial.sql) database table then run the following job with correct environment variables.
+
 ```sh
 npx @socialgouv/matomo-postgres
 ```
@@ -31,7 +35,7 @@ export MATOMO_URL=
 export MATOMO_SITE=
 export MATOMO_KEY=
 export DESTINATION_TABLE= # optional
-export STARTDATE= # optional 
+export STARTDATE= # optional
 export OFFSET= # optional
 export PGDATABASE=postgres://postgres:postgres@127.0.0.1:5455/postgres
 yarn start

docker-compose.yml

@@ -1,10 +1,18 @@
 version: "3.0"
 services:
   postgres:
-    image: postgres:11
+    build:
+      context: ./docker
+      dockerfile: ./Dockerfile
+    volumes:
+      - postgres_data:/var/lib/postgresql/data
     environment:
       POSTGRES_PASSWORD: postgres
       POSTGRES_USERNAME: postgres
       TZ: "Europe/Paris"
     ports:
       - 5455:5432
+
+
+volumes:
+  postgres_data:

docker/Dockerfile

@@ -0,0 +1,43 @@
+FROM postgres:13-alpine
+
+ENV PG_PARTMAN_VERSION v4.7.0
+
+# Install pg_jobmon
+RUN set -ex \
+    \
+    && apk add --no-cache --virtual .fetch-deps \
+    ca-certificates \
+    openssl \
+    tar \
+    \
+    && apk add --no-cache --virtual .build-deps \
+    autoconf \
+    automake \
+    g++ \
+    clang \
+    llvm \
+    libtool \
+    libxml2-dev \
+    make \
+    perl
+# Install pg_partman
+RUN set -ex \
+    && wget -O pg_partman.tar.gz "https://github.com/pgpartman/pg_partman/archive/$PG_PARTMAN_VERSION.tar.gz" \
+    && mkdir -p /usr/src/pg_partman \
+    && tar \
+    --extract \
+    --file pg_partman.tar.gz \
+    --directory /usr/src/pg_partman \
+    --strip-components 1 \
+    && rm pg_partman.tar.gz \
+    && cd /usr/src/pg_partman \
+    && make \
+    && make install \
+    && cd / \
+    && rm -rf /usr/src/pg_partman \
+    && apk del .fetch-deps .build-deps
+
+# Copy the init script
+# The Docker Postgres initd script will run anything 
+# in the directory /docker-entrypoint-initdb.d
+COPY initdb.sh /docker-entrypoint-initdb.d/initdb.sh

docker/initdb.sh

@@ -0,0 +1,13 @@
+#!/bin/bash -e
+
+echo "Creating partman extension"
+psql -v ON_ERROR_STOP=1 --username "$POSTGRES_USER" --dbname "$POSTGRES_DB" <<-EOSQL
+    CREATE SCHEMA partman;
+    CREATE EXTENSION pg_partman SCHEMA partman;
+EOSQL
+
+echo "ADDING pg_partman_bgw TO postgresql.conf"
+echo "shared_preload_libraries = 'pg_partman_bgw'" >> $PGDATA/postgresql.conf
+echo "pg_partman_bgw.interval = 3600" >> $PGDATA/postgresql.conf
+echo "pg_partman_bgw.role = '$POSTGRES_USER'" >> $PGDATA/postgresql.conf
+echo "pg_partman_bgw.dbname = '$POSTGRES_DB'" >> $PGDATA/postgresql.conf

initial.sql

@@ -0,0 +1,73 @@
+-- converting existing matomo table to partitioned witg pg_partman
+-- usage : ON_ERROR_STOP=1 psql < partition.sql
+
+--- pg_partman setup
+
+CREATE SCHEMA IF NOT EXISTS partman;
+CREATE EXTENSION IF NOT EXISTS pg_partman SCHEMA partman;
+
+--- backup and recreate a new partionned matomo table
+
+CREATE TABLE IF NOT EXISTS matomo_tmp as (select * from matomo);
+
+ALTER TABLE IF EXISTS matomo RENAME TO matomo_backup;
+
+CREATE TABLE IF NOT EXISTS matomo
+(
+    idsite                      text,
+    idvisit                     text,
+    actions                     text,
+    country                     text,
+    region                      text,
+    city                        text,
+    operatingsystemname         text,
+    devicemodel                 text,
+    devicebrand                 text,
+    visitduration               text,
+    dayssincefirstvisit         text,
+    visitortype                 text,
+    sitename                    text,
+    userid                      text,
+    serverdateprettyfirstaction date,
+    action_id                   text,
+    action_type                 text,
+    action_eventcategory        text,
+    action_eventaction          text,
+    action_eventname            text,
+    action_eventvalue           decimal,
+    action_timespent            text,
+    action_timestamp            timestamp with time zone DEFAULT now(),
+    usercustomproperties        json,
+    usercustomdimensions        json,
+    dimension1                  text,
+    dimension2                  text,
+    dimension3                  text,
+    dimension4                  text,
+    dimension5                  text,
+    dimension6                  text,
+    dimension7                  text,
+    dimension8                  text,
+    dimension9                  text,
+    dimension10                 text,
+    action_url                  text,
+    sitesearchkeyword           text,
+    action_title                text
+) PARTITION BY RANGE (action_timestamp);
+
+ALTER TABLE IF EXISTS matomo ADD CONSTRAINT unique_action_id UNIQUE (action_id, action_timestamp);
+ALTER TABLE IF EXISTS matomo ALTER COLUMN action_eventvalue TYPE decimal USING action_eventvalue::decimal;
+CREATE INDEX IF NOT EXISTS idx_action_timestamp_matomo ON matomo (action_timestamp);
+CREATE INDEX IF NOT EXISTS idx_idvisit_matomo ON matomo(idvisit);
+CREATE INDEX IF NOT EXISTS idx_action_eventcategory_matomo ON matomo(action_eventcategory);
+CREATE INDEX IF NOT EXISTS idx_action_type_matomo ON matomo(action_type);
+CREATE INDEX IF NOT EXISTS idx_action_eventaction_matomo ON matomo(action_eventaction);
+
+SELECT partman.create_parent('public.matomo', 'action_timestamp', 'native', 'monthly');
+
+-- Import des données depuis la table standard vers la table partitionnée
+CALL partman.partition_data_proc('public.matomo', p_source_table := 'public.matomo_tmp', p_order:= 'DESC');
+
+VACUUM ANALYZE public.matomo;
+
+DROP TABLE if exists matomo_tmp;
+

src/__tests__/__snapshots__/index.test.js.snap

@@ -2,8 +2,12 @@
 
 exports[`run: should create table 1`] = `
 Array [
-  "CREATE TABLE IF NOT EXISTS matomo
-  (
+  "
+
+CREATE SCHEMA IF NOT EXISTS partman;
+CREATE EXTENSION IF NOT EXISTS pg_partman SCHEMA partman;
+CREATE TABLE IF NOT EXISTS matomo
+(
     idsite                      text,
     idvisit                     text,
     actions                     text,
@@ -19,14 +23,14 @@ Array [
     sitename                    text,
     userid                      text,
     serverdateprettyfirstaction date,
-    action_id                   text UNIQUE,
+    action_id                   text,
     action_type                 text,
     action_eventcategory        text,
     action_eventaction          text,
     action_eventname            text,
     action_eventvalue           decimal,
     action_timespent            text,
-    action_timestamp            timestamp with time zone,
+    action_timestamp            timestamp with time zone DEFAULT now(),
     usercustomproperties        json,
     usercustomdimensions        json,
     dimension1                  text,
@@ -42,7 +46,8 @@ Array [
     action_url                  text,
     sitesearchkeyword           text,
     action_title                text
-)",
+) PARTITION BY RANGE (action_timestamp);
+",
   Array [],
 ]
 `;

src/__tests__/index.test.js

@@ -12,7 +12,7 @@ const matomoVisit = require("./visit.json");
 
 const run = require("../index");
 
-const NB_REQUEST_TO_INIT_DB = 21; // Number of query to init DB (createTable.js)
+const NB_REQUEST_TO_INIT_DB = 1; // Number of query to init DB (createTable.js)
 const TEST_DATE = new Date();
 
 // @ts-ignore
@@ -86,11 +86,6 @@ test("run: should fetch the latest event date if no date provided", async () =>
 
   // check db queries
   expect(mock_pgQuery.mock.calls[NB_REQUEST_TO_INIT_DB][0]).toEqual(
-    // call 0 is create table
-    // call 1 is add column usercustomdimension
-    // call 2 is add column action_url
-    // ...
-    //
     "select action_timestamp from matomo order by action_timestamp desc limit 1"
   );
 });

src/createTable.js

@@ -8,8 +8,12 @@ const { DESTINATION_TABLE } = require("./config");
  */
 async function createTable(client) {
   const table = client.escapeIdentifier(DESTINATION_TABLE);
-  const text = `CREATE TABLE IF NOT EXISTS ${table}
-  (
+  const text = `
+
+CREATE SCHEMA IF NOT EXISTS partman;
+CREATE EXTENSION IF NOT EXISTS pg_partman SCHEMA partman;
+CREATE TABLE IF NOT EXISTS ${table}
+(
     idsite                      text,
     idvisit                     text,
     actions                     text,
@@ -25,14 +29,14 @@ async function createTable(client) {
     sitename                    text,
     userid                      text,
     serverdateprettyfirstaction date,
-    action_id                   text UNIQUE,
+    action_id                   text,
     action_type                 text,
     action_eventcategory        text,
     action_eventaction          text,
     action_eventname            text,
     action_eventvalue           decimal,
     action_timespent            text,
-    action_timestamp            timestamp with time zone,
+    action_timestamp            timestamp with time zone DEFAULT now(),
     usercustomproperties        json,
     usercustomdimensions        json,
     dimension1                  text,
@@ -48,37 +52,12 @@ async function createTable(client) {
     action_url                  text,
     sitesearchkeyword           text,
     action_title                text
-)`;
+) PARTITION BY RANGE (action_timestamp);
+`;
 
   await client.query(text, []);
 
-  const migrations = [
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "usercustomdimensions" json;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "action_url" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "sitesearchkeyword" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "action_title" text;`,
-    `ALTER TABLE IF EXISTS ${table} ALTER COLUMN action_eventvalue TYPE decimal USING action_eventvalue::decimal;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension1" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension2" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension3" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension4" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension5" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension6" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension7" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension8" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension9" text;`,
-    `ALTER TABLE IF EXISTS ${table} ADD COLUMN IF NOT EXISTS "dimension10" text;`,
-    `CREATE INDEX IF NOT EXISTS idx_action_timestamp ON ${table} (action_timestamp);`,
-    `CREATE INDEX IF NOT EXISTS idx_idvisit ON ${table}(idvisit);`,
-    `CREATE INDEX IF NOT EXISTS idx_action_eventcategory ON ${table}(action_eventcategory);`,
-    `CREATE INDEX IF NOT EXISTS idx_action_type ON ${table}(action_type);`,
-    `CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_action_eventaction ON ${table}(action_eventaction);`,
-  ];
-
-  // --------------------------------------------- //
-  // If you add new query: Don't forget to update  //
-  // const `NB_REQUEST_TO_INIT_DB` (index.test.js) //
-  // --------------------------------------------- //
+  const migrations = [];
 
   for (const query of migrations) {
     await client.query(query, []);