Topo sur les premières versions publique de MySQL

juillet 19, 2017

j’ai régulièrement l’opportunité de rencontrer les utilisateurs des produits MySQL, et je suis toujours un peu surpris de voir des applications critiques qui tournent sur des versions pas vraiment récente (pour employer un euphémisme) 🙂

La bonne nouvelle est que manifestement les anciennes versions de MySQL sont suffisamment stables et performantes pour faire tourner du business moderne. Cependant, et ce même si je comprend bien qu’il est parfois pertinent de figer toutes les couches d’une architecture, il est souvent dommage de ne pas profiter des dernières améliorations d’un point de vue, performance, stabilité, sécurité et bien entendu des nouvelles fonctionnalités de la dernière GA:

La notion de temps étant relative, je te propose un petit jeu de mise en perspective des dates des premières versions publique (non GA), une sorte de blind-test MySQL…

MySQL 3.22 | 1998

Voiture la plus vendue en France : Renault Clio II (phase 1)
Meilleure vente de Singles en France : Garou, Daniel Lavoie et Patrick Fiori, Belle
Evénement : 12 juillet 1998, la France bat le Brésil lors de la finale de la coupe du monde de football 3 buts à 0

Sources:
http://dev.cs.ovgu.de/db/mysql/News-3.22.x.html
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=18
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_1998
https://fr.wikipedia.org/wiki/1998_en_France

MySQL 3.23 | 1999

Voiture la plus vendue en France : Renault Clio II (phase 1)
Meilleure vente de Singles en France : Lou Bega, Mambo No. 5
Evénement : 11 août 1999, une éclipse totale de Soleil se produit sur le Nord de la France

Sources:
http://mysql.localhost.net.ar/doc/refman/4.1/en/news-3-23-x.html
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=18
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_1999
https://fr.wikipedia.org/wiki/1999_en_France

MySQL 4.0 | 2001

Voiture la plus vendue en France : Peugeot 206 (phase 1)
Meilleure vente de Singles en France : Star Academy 1, La musique
Evénement : 27 juin, fin de la conscription. L’armée française est désormais entièrement professionnelle

Sources:
http://mysql.localhost.net.ar/doc/refman/4.1/en/news-4-0-x.html
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=19
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2001
https://fr.wikipedia.org/wiki/2001_en_France

MySQL 4.1 | 2003

Voiture la plus vendue en France : Renault Clio II (phase 2)
Meilleure vente de Singles en France : DJ BoBo, Chihuahua
Evénement : 30 mai, dernier vol du Concorde d’Air France entre Paris et New York

Sources:
http://mysql.localhost.net.ar/doc/refman/4.1/en/news-4-1-x.html
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=20
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2003
https://fr.wikipedia.org/wiki/2003_en_France

MySQL 5.0 | 2003

Voiture la plus vendue en France : Renault Clio II (phase 2)
Meilleure vente de Singles en France : DJ BoBo, Chihuahua
Evénement : 30 mai, dernier vol du Concorde d’Air France entre Paris et New York

Sources:
http://ftp.nchu.edu.tw/MySQL/doc/refman/5.0/en/news-5-0-x.html
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=20
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2003
https://fr.wikipedia.org/wiki/2003_en_France

MySQL 5.1 | 2005

Voiture la plus vendue en France : Renault Clio III (phase 1)
Meilleure vente de Singles en France : Ilona Mitrecey, Un monde parfait
Evénement : 31 mars, lancement de la TNT (14 chaînes gratuites).

Sources:
http://ftp.nchu.edu.tw/MySQL/doc/refman/5.1/en/news-5-1-x.html
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=22
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2005
https://fr.wikipedia.org/wiki/2005_en_France

MySQL 5.5 | 2009

Voiture la plus vendue en France : Peugeot 207
Meilleure vente de Singles en France : Helmut Fritz, Ça m’énerve
Evénement : 15 avril, entrée en vigueur des nouvelles plaques d’immatriculations sur les véhicules neufs. La nouvelle numérotation est attribuée à vie pour chaque véhicule et comporte deux lettres, un tiret, trois chiffres, un tiret et deux lettres. Un numéro de département au choix ainsi que le logo de la région correspondante est apposée sur la droite

Sources:
https://dev.mysql.com/doc/relnotes/mysql/5.5/en/
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=23
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2009
https://fr.wikipedia.org/wiki/2009_en_France

MySQL 5.6 | 2011

Voiture la plus vendue en France : Renault Clio III (phase 2)
Meilleure vente de Singles en France : Israel Kamakawiwo’ole : Over the Rainbow
Evénement : 14 mai : arrestation à New-York de Dominique Strauss-Kahn, directeur général du FMI, dans le cadre d’une accusation de crimes sexuels à l’encontre de Nafissatou Diallo

Sources:
https://dev.mysql.com/doc/relnotes/mysql/5.6/en/
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=24
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2011
https://fr.wikipedia.org/wiki/2011_en_France

MySQL 5.7 | 2013

Voiture la plus vendue en France : Renault Clio IV
Meilleure vente de Singles en France : Daft Punk, Get Lucky
Evénement : 9 février, la découverte de viande de cheval camouflée dans des préparations surgelées réputées pur bœuf renforce les soupçons de tromperie sur la traçabilité agroalimentaire à l’échelle de la communauté économique européenne

Sources:
https://dev.mysql.com/doc/relnotes/mysql/5.7/en/
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=25
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2013
https://fr.wikipedia.org/wiki/2013_en_France

MySQL 8.0 | 2016

Voiture la plus vendue en France : Renault Clio IV (phase 2)
Meilleure vente de Singles en France : M Pokora, Cette Année la
Evénement : 1er janvier, entrée en vigueur du nouveau découpage des régions

Sources:
https://dev.mysql.com/doc/relnotes/mysql/8.0/en/
http://www.auto-moto.com/occasion/souvenirs-souvenirs/voiture-la-plus-vendue-par-annee-les-statistiques-de-1947-a-nos-jours-55637.html#item=26
https://fr.wikipedia.org/wiki/Liste_des_titres_musicaux_num%C3%A9ro_un_en_France_en_2016
https://fr.wikipedia.org/wiki/2016_en_France

Sakila in Norway

Thanks for using MySQL!

Filed under:
Astuce, MySQL by Olivier DASINI

PHP Tour 2017 – Slides MySQL InnoDB Cluster

mai 26, 2017

Olivier DASINI aka @freshdaz La dernière édition du PHP Tour s’est déroulée les 18 et 19 mai 2017 à Nantes. Ce que j’en garde : un très bon cru, de bien belles rencontres, de bonnes bières ainsi qu’une excellente organisation (merci l’AFUP).

J’ai également eu l’opportunité de présenter MySQL InnoDB Cluster, la nouvelle solution native de haute disponibilité de MySQL :

MySQL InnoDB Cluster – A complete High Availability solution for MySQL from Olivier DASINI

Vidéo de la conférence : https://youtu.be/Y48cbFqd5QA

Les photos sont disponibles sur le groupe Flickr du PHP Tour 2017 Nantes.

Les vidéos des conférences sont disponibles sur le site de l’AFUP.

Thanks for using MySQL!

Filed under:
Conférence, Group Replication, Haute Disponibilité, InnoDB Cluster, PHP, Présentation by Olivier DASINI

Tutoriel – Déployer MySQL innoDB Cluster

mai 11, 2017

Dans les épisodes précédents on a vu comment déployer « manuellement » MySQL Group Replication, comprendre et tester MySQL InnoDB Cluster ainsi que comment gérer aisément un cluster Group Replication déja déployé avec MySQL Shell.

Aujourd’hui, dans la série Haute Disponibilité avec MySQL on va voir comment déployer et gérer un cluster from scratch , sous la forme d’un tutoriel, grâce à la solution tout en un : MySQL InnoDB Cluster.

Note:
L’article traite de MySQL InnoDB Cluster, HA natif de MySQL Server (plugin GRoup Replication) avec pour moteur de stockage InnoDB, solution à ne pas confondre avec MySQL NDB Cluster (moteur de stockage NDB).

Le contexte

3 instances MySQL autonomes, <spoil altert> qui vont grâce au plugin MySQL Group Replication se transformer en une base de données distribuée.</spoil altert>

Instance MySQL 1 : 192.168.1.22; Nœud numéro 1 du cluster
Instance MySQL 2 : 192.168.1.50; Nœud numéro 2 du cluster
Instance MySQL 3 : 192.168.1.48; Nœud numéro 3 du cluster

1 instance applicative : 192.168.1.11; MySQL Router + mon application.

MySQL Shell est installé sur toutes les instances.

En ce qui concerne les versions des logiciels:

Avec un schéma sa donne (à peu près) ça :

MySQL InnoDB Cluster PoC Architecture

Vérifier la configuration des instances

La première étape consiste à s’assurer que les instances MySQL sont correctement configurées pour l’utilisation de MySQL Group Replication, la couche haute disponibilité de notre architecture. A noter qu’il est préférable de provisionner ses instances déjà correctement configurées (comme détaillé dans cet article) pour MySQL Group Replication.

La vérification de la configuration se fait grâce à MySQL Shell et la méthode checkInstanceConfiguration() :

$ mysqlsh
Welcome to MySQL Shell 1.0.9

Copyright (c) 2016, 2017, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.



mysql-js> dba.checkInstanceConfiguration('root@192.168.1.22:3306')
Please provide the password for 'root@192.168.1.22:3306': 
Validating instance...

The instance '192.168.1.22:3306' is not valid for Cluster usage.

The following issues were encountered:

 - Some configuration options need to be fixed.

+----------------------------------+---------------+----------------+--------------------------------------------------+
| Variable                         | Current Value | Required Value | Note                                             |
+----------------------------------+---------------+----------------+--------------------------------------------------+
| binlog_checksum                  | CRC32         | NONE           | Update the server variable or restart the server |
| enforce_gtid_consistency         | OFF           | ON             | Restart the server                               |
| gtid_mode                        | OFF           | ON             | Restart the server                               |
| log_bin                          | 0             | 1              | Restart the server                               |
| log_slave_updates                | 0             | ON             | Restart the server                               |
| master_info_repository           | FILE          | TABLE          | Restart the server                               |
| relay_log_info_repository        | FILE          | TABLE          | Restart the server                               |
| transaction_write_set_extraction | OFF           | XXHASH64       | Restart the server                               |
+----------------------------------+---------------+----------------+--------------------------------------------------+


Please fix these issues , restart the serverand try again.

{
    "config_errors": [
        {
            "action": "server_update", 
            "current": "CRC32", 
            "option": "binlog_checksum", 
            "required": "NONE"
        },
        {
            "action": "restart", 
            "current": "OFF", 
            "option": "enforce_gtid_consistency", 
            "required": "ON"
        },
        {
            "action": "restart", 
            "current": "OFF", 
            "option": "gtid_mode", 
            "required": "ON"
        },
        {
            "action": "restart", 
            "current": "0", 
            "option": "log_bin", 
            "required": "1"
        },
        {
            "action": "restart", 
            "current": "0", 
            "option": "log_slave_updates", 
            "required": "ON"
        },
        {
            "action": "restart", 
            "current": "FILE", 
            "option": "master_info_repository", 
            "required": "TABLE"
        },
        {
            "action": "restart", 
            "current": "FILE", 
            "option": "relay_log_info_repository", 
            "required": "TABLE"
        },
        {
            "action": "restart", 
            "current": "OFF", 
            "option": "transaction_write_set_extraction", 
            "required": "XXHASH64"
        }
    ], 
    "errors": [], 
    "restart_required": true, 
    "status": "error"
}

$ mysqlsh

Welcome to MySQL Shell 1.0.9

Oracle is a registered trademark of Oracle Corporation and/or its

affiliates. Other names may be trademarks of their respective

owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.

mysql-js> dba.checkInstanceConfiguration('root@192.168.1.22:3306')

Please provide the password for 'root@192.168.1.22:3306':

Validating instance...

The instance '192.168.1.22:3306' is not valid for Cluster usage.

The following issues were encountered:

- Some configuration options need to be fixed.

+----------------------------------+---------------+----------------+--------------------------------------------------+

+----------------------------------+---------------+----------------+--------------------------------------------------+

| enforce_gtid_consistency | OFF | ON | Restart the server |

| gtid_mode | OFF | ON | Restart the server |

| log_bin | 0 | 1 | Restart the server |

| log_slave_updates | 0 | ON | Restart the server |

+----------------------------------+---------------+----------------+--------------------------------------------------+

Please fix these issues , restart the serverand try again.

{

"config_errors": [

{

"action": "server_update",

"current": "CRC32",

"option": "binlog_checksum",

"required": "NONE"

{

"action": "restart",

"current": "OFF",

"option": "enforce_gtid_consistency",

"required": "ON"

{

"action": "restart",

"current": "OFF",

"option": "gtid_mode",

"required": "ON"

{

"action": "restart",

"current": "0",

"option": "log_bin",

"required": "1"

{

"action": "restart",

"current": "0",

"option": "log_slave_updates",

"required": "ON"

{

"action": "restart",

"current": "FILE",

"option": "master_info_repository",

"required": "TABLE"

{

"action": "restart",

"current": "FILE",

"option": "relay_log_info_repository",

"required": "TABLE"

{

"action": "restart",

"current": "OFF",

"option": "transaction_write_set_extraction",

"required": "XXHASH64"

}

"errors": [],

"restart_required": true,

"status": "error"

}

Dans mon cas, avec l’installation par défaut sous Ubuntu, niveau configuration j’ai tout à faire 🙂

La méthode renvoie un document JSON (pratique pour l’automatisation) avec la liste des tâches à effectuer pour être conforme… Configurons donc !

J’ai deux solutions :

1/ je prépare mes instances « manuellement » (cette tâche peut bien évidemment s’automatiser) comme expliqué dans l’article comment configurer un groupe.
2/ je me connecte à chaque instance en local, et j’utilise la méthode : configureLocalInstance()

Soyons fou ! allons y pour la méthode 2 :

mysql-js> dba.configureLocalInstance('root@192.168.1.22:3306')
Dba.configureLocalInstance: This function only works with local instances (RuntimeError)

1 2	mysql-js> dba.configureLocalInstance('root@192.168.1.22:3306') Dba.configureLocalInstance: This function only works with local instances (RuntimeError)

Ouppss!!! dba.configureLocalInstance ne fonctionne qu’en local, c’est-à-dire, si je suis connecté sur la machine hôte de l’instance MySQL (ce qui est une bonne chose). Du coup après m’être connecté à l’hôte 192.168.1.22 :

sudo mysqlsh
Welcome to MySQL Shell 1.0.9
...

mysql-js> dba.configureLocalInstance('root@localhost:3306')
Please provide the password for 'root@localhost:3306': 

Detecting the configuration file...
Found configuration file at standard location: /etc/mysql/mysql.conf.d/mysqld.cnf
Do you want to modify this file? [Y|n]: Y
Validating instance...

The configuration has been updated but it is required to restart the server.

{
    "config_errors": [
        {
            "action": "restart", 
            "current": "OFF", 
            "option": "enforce_gtid_consistency", 
            "required": "ON"
        },
        {
            "action": "restart", 
            "current": "OFF", 
            "option": "gtid_mode", 
            "required": "ON"
        },
        {
            "action": "restart", 
            "current": "0", 
            "option": "log_bin", 
            "required": "1"
        },
        {
            "action": "restart", 
            "current": "0", 
            "option": "log_slave_updates", 
            "required": "ON"
        },
        {
            "action": "restart", 
            "current": "FILE", 
            "option": "master_info_repository", 
            "required": "TABLE"
        },
        {
            "action": "restart", 
            "current": "FILE", 
            "option": "relay_log_info_repository", 
            "required": "TABLE"
        },
        {
            "action": "restart", 
            "current": "OFF", 
            "option": "transaction_write_set_extraction", 
            "required": "XXHASH64"
        }
    ], 
    "errors": [], 
    "restart_required": true, 
    "status": "error"
}

sudo mysqlsh

Welcome to MySQL Shell 1.0.9

...

mysql-js> dba.configureLocalInstance('root@localhost:3306')

Please provide the password for 'root@localhost:3306':

Detecting the configuration file...

Found configuration file at standard location: /etc/mysql/mysql.conf.d/mysqld.cnf

Do you want to modify this file? [Y|n]: Y

Validating instance...

The configuration has been updated but it is required to restart the server.

{

"config_errors": [

{

"action": "restart",

"current": "OFF",

"option": "enforce_gtid_consistency",

"required": "ON"

{

"action": "restart",

"current": "OFF",

"option": "gtid_mode",

"required": "ON"

{

"action": "restart",

"current": "0",

"option": "log_bin",

"required": "1"

{

"action": "restart",

"current": "0",

"option": "log_slave_updates",

"required": "ON"

{

"action": "restart",

"current": "FILE",

"option": "master_info_repository",

"required": "TABLE"

{

"action": "restart",

"current": "FILE",

"option": "relay_log_info_repository",

"required": "TABLE"

{

"action": "restart",

"current": "OFF",

"option": "transaction_write_set_extraction",

"required": "XXHASH64"

}

"errors": [],

"restart_required": true,

"status": "error"

}

Note: Assure toi d’avoir les droits nécessaires pour mettre à jour le fichier de configuration de MySQL.

Les informations ajoutées dans le fichier de configuration se trouvent en fin de fichier :

# cat /etc/mysql/mysql.conf.d/mysqld.cnf

[mysqld]
... <data previously in the file> ...

log_slave_updates = ON
server_id = 2323701546
relay_log_info_repository = TABLE
master_info_repository = TABLE
transaction_write_set_extraction = XXHASH64
binlog_format = ROW
disabled_storage_engines = MyISAM,BLACKHOLE,FEDERATED,CSV,ARCHIVE
report_port = 3306
binlog_checksum = NONE
enforce_gtid_consistency = ON
log_bin
gtid_mode = ON

# cat /etc/mysql/mysql.conf.d/mysqld.cnf

[mysqld]

... <data previously in the file> ...

log_slave_updates = ON

server_id = 2323701546

relay_log_info_repository = TABLE

master_info_repository = TABLE

transaction_write_set_extraction = XXHASH64

binlog_format = ROW

disabled_storage_engines = MyISAM,BLACKHOLE,FEDERATED,CSV,ARCHIVE

report_port = 3306

binlog_checksum = NONE

enforce_gtid_consistency = ON

log_bin

gtid_mode = ON

192.168.1.22 est configurée. Après redémarrage de l’instance MySQL, la sortie de checkInstanceConfiguration est beaucoup moins anxiogène :

mysql-js> dba.checkInstanceConfiguration('root@192.168.1.22:3306')
Please provide the password for 'root@192.168.1.22:3306': 
Validating instance...

The instance '192.168.1.22:3306' is valid for Cluster usage
{
    "status": "ok"
}

mysql-js> dba.checkInstanceConfiguration('root@192.168.1.22:3306')

Please provide the password for 'root@192.168.1.22:3306':

Validating instance...

The instance '192.168.1.22:3306' is valid for Cluster usage

{

"status": "ok"

}

OK ! Le membre est prêt pour faire parti d’un groupe.

La même procédure doit être appliquée sur les autres instances MySQL.

…

mysql-js> dba.checkInstanceConfiguration('root@192.168.1.50:3306')
Please provide the password for 'root@192.168.1.50:3306': 
Validating instance...

The instance '192.168.1.50:3306' is valid for Cluster usage
{
    "status": "ok"
}
mysql-js> dba.checkInstanceConfiguration('root@192.168.1.48:3306')
Please provide the password for 'root@192.168.1.48:3306': 
Validating instance...

The instance '192.168.1.48:3306' is valid for Cluster usage
{
    "status": "ok"
}

mysql-js> dba.checkInstanceConfiguration('root@192.168.1.50:3306')

Please provide the password for 'root@192.168.1.50:3306':

Validating instance...

The instance '192.168.1.50:3306' is valid for Cluster usage

{

"status": "ok"

}

mysql-js> dba.checkInstanceConfiguration('root@192.168.1.48:3306')

Please provide the password for 'root@192.168.1.48:3306':

Validating instance...

The instance '192.168.1.48:3306' is valid for Cluster usage

{

"status": "ok"

}

Créer le cluster

Une fois les 3 instances correctement configurées, l’étape suivante consiste à créer le cluster avec createCluster. Cette méthode va être jouée sur le premier membre, l’instance MySQL sur 192.168.1.22, elle va permettre de bootstrapper le cluster:

mysql-js> \c root@192.168.1.22
Creating a Session to 'root@192.168.1.22'
Enter password: 
Classic Session successfully established. No default schema selected.


mysql-js> var cluster = dba.createCluster('pocCluster', {ipWhitelist: "192.168.1.0/16"});
A new InnoDB cluster will be created on instance 'root@192.168.1.22:3306'.

Creating InnoDB cluster 'pocCluster' on 'root@192.168.1.22:3306'...
Adding Seed Instance...

Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.
At least 3 instances are needed for the cluster to be able to withstand up to
one server failure.

mysql-js> \c root@192.168.1.22

Creating a Session to 'root@192.168.1.22'

Enter password:

Classic Session successfully established. No default schema selected.

mysql-js> var cluster = dba.createCluster('pocCluster', {ipWhitelist: "192.168.1.0/16"});

A new InnoDB cluster will be created on instance 'root@192.168.1.22:3306'.

Creating InnoDB cluster 'pocCluster' on 'root@192.168.1.22:3306'...

Adding Seed Instance...

Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.

At least 3 instances are needed for the cluster to be able to withstand up to

one server failure.

createCluster() prend comme paramètre le nom du cluster (pocCluster). Je peux lui passer également quelques information optionnelles comme la whitelist.

On peut vérifier l’état du nœud dans le cluster avec status() :

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Notes : Assure toi que ton DNS (ou /etc/hosts) est correctement configuré, sinon tu vas avoir des soucis de connections…

L’ajouts des nœuds suivant se fait avec addInstance(), il est néanmoins conseillé d’exécuter checkInstanceState() au préalable pour s’assurer de la compatibilité des GTID sets :

Nœud 2

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')
Please provide the password for 'root@192.168.1.50:3306': 
Analyzing the instance replication state...

The instance '192.168.1.50:3306' is valid for the cluster.
The instance is new to Group Replication.

{
    "reason": "new", 
    "state": "ok"
}


mysql-js> cluster.addInstance("root@192.168.1.50:3306", {ipWhitelist: "192.168.1.0/16"})
A new instance will be added to the InnoDB cluster. Depending on the amount of
data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@192.168.1.50:3306': 
Adding instance to the cluster ...

The instance 'root@192.168.1.50:3306' was successfully added to the cluster.


mysql-js> cluster.status();
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')

Please provide the password for 'root@192.168.1.50:3306':

Analyzing the instance replication state...

The instance '192.168.1.50:3306' is valid for the cluster.

The instance is new to Group Replication.

{

"reason": "new",

"state": "ok"

}

mysql-js> cluster.addInstance("root@192.168.1.50:3306", {ipWhitelist: "192.168.1.0/16"})

A new instance will be added to the InnoDB cluster. Depending on the amount of

data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@192.168.1.50:3306':

Adding instance to the cluster ...

The instance 'root@192.168.1.50:3306' was successfully added to the cluster.

mysql-js> cluster.status();

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Au cas où l’instance ajoutée contient plus de transactions que le groupe checkInstanceState le fait savoir :

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')
Please provide the password for 'root@192.168.1.50:3306': 
Analyzing the instance replication state...

The instance '192.168.1.50:3306' is invalid for the cluster.
The instance contains additional transactions in relation to the cluster.

{
    "reason": "diverged", 
    "state": "error"
}

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')

Please provide the password for 'root@192.168.1.50:3306':

Analyzing the instance replication state...

The instance '192.168.1.50:3306' is invalid for the cluster.

The instance contains additional transactions in relation to the cluster.

{

"reason": "diverged",

"state": "error"

}

En fonction du contexte, il faut alors soit restaurer une sauvegarde d’un membre du cluster sur l’instance problématique (celle qui diverge) ou alors si tu sais ce que tu fais, une synchronisation des GTIDs est toujours possible, voir un reset master.

Nœud 3

mysql-js> cluster.checkInstanceState('root@192.168.1.48:3306')
Please provide the password for 'root@192.168.1.48:3306': 
Analyzing the instance replication state...

The instance '192.168.1.48:3306' is valid for the cluster.
The instance is new to Group Replication.

{
    "reason": "new", 
    "state": "ok"
}


mysql-js> cluster.addInstance("root@192.168.1.48:3306", {ipWhitelist: "192.168.1.0/16"})
A new instance will be added to the InnoDB cluster. Depending on the amount of
data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@192.168.1.48:3306': 
Adding instance to the cluster ...

The instance 'root@192.168.1.48:3306' was successfully added to the cluster.


mysql-js> cluster.status();
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.checkInstanceState('root@192.168.1.48:3306')

Please provide the password for 'root@192.168.1.48:3306':

Analyzing the instance replication state...

The instance '192.168.1.48:3306' is valid for the cluster.

The instance is new to Group Replication.

{

"reason": "new",

"state": "ok"

}

mysql-js> cluster.addInstance("root@192.168.1.48:3306", {ipWhitelist: "192.168.1.0/16"})

A new instance will be added to the InnoDB cluster. Depending on the amount of

data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@192.168.1.48:3306':

Adding instance to the cluster ...

The instance 'root@192.168.1.48:3306' was successfully added to the cluster.

mysql-js> cluster.status();

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Et voilà!

Cluster MySQL Group Replication de 3 nœuds déployé grâce à MySQL Shell !

Nœud 1 : 192.168.1.22 : Primaire (lecture/écriture)
Nœud 2 : 192.168.1.50 : Secondaire (lecture seule)
Nœud 3 : 192.168.1.48 : Secondaire (lecture seule)

Pour rendre la configuration persistante ie l’écrire dans le fichier de configuration, il faut exécuter, après que le noeud soit configuré, la méthode dba.configureLocalInstance() :

daz@192.168.1.50:~$ grep -c group_replication /etc/mysql/mysql.conf.d/mysqld.cnf
0


daz@192.168.1.50:~$ sudo mysqlsh --uri=root@localhost
Creating a Session to 'root@localhost'
Enter password: 
Classic Session successfully established. No default schema selected.
Welcome to MySQL Shell 1.0.9
...


mysql-js> dba.configureLocalInstance("root@localhost:3306")  
Please provide the password for 'root@localhost:3306': 

Detecting the configuration file...
Found configuration file at standard location: /etc/mysql/mysql.conf.d/mysqld.cnf
Do you want to modify this file? [Y|n]: Y
Validating instance...

The instance 'localhost:3306' is valid for Cluster usage
You can now use it in an InnoDB Cluster.

{
    "status": "ok"
}



daz@192.168.1.50:~$ grep -c group_replication /etc/mysql/my.cnf
32

daz@192.168.1.50:~$ grep -c group_replication /etc/mysql/mysql.conf.d/mysqld.cnf

daz@192.168.1.50:~$ sudo mysqlsh --uri=root@localhost

Creating a Session to 'root@localhost'

Enter password:

Classic Session successfully established. No default schema selected.

Welcome to MySQL Shell 1.0.9

...

mysql-js> dba.configureLocalInstance("root@localhost:3306")

Please provide the password for 'root@localhost:3306':

Detecting the configuration file...

Found configuration file at standard location: /etc/mysql/mysql.conf.d/mysqld.cnf

Do you want to modify this file? [Y|n]: Y

Validating instance...

The instance 'localhost:3306' is valid for Cluster usage

You can now use it in an InnoDB Cluster.

{

"status": "ok"

}

daz@192.168.1.50:~$ grep -c group_replication /etc/mysql/my.cnf

A noter que cette opération ne peut se faire qu’en local.

La suite ?

Configuration de MySQL Router

Les recommandations de MySQL sont d’installer MySQL Router sur la machine hôte de l’application, je vais donc l’installer sur la machine 192.168.1.11.

Bootstrap MySQL Router

daz@192.168.1.11 ~ $ mysqlrouter --bootstrap root@192.168.1.22:3306 --directory RouterPoC
Please enter MySQL password for root: 

Bootstrapping MySQL Router instance at /home/daz/RouterPoC...
MySQL Router  has now been configured for the InnoDB cluster 'pocCluster'.

The following connection information can be used to connect to the cluster.

Classic MySQL protocol connections to cluster 'pocCluster':
- Read/Write Connections: localhost:6446
- Read/Only Connections: localhost:6447

X protocol connections to cluster 'pocCluster':
- Read/Write Connections: localhost:64460
- Read/Only Connections: localhost:64470

daz@192.168.1.11 ~ $ mysqlrouter --bootstrap root@192.168.1.22:3306 --directory RouterPoC

Please enter MySQL password for root:

Bootstrapping MySQL Router instance at /home/daz/RouterPoC...

MySQL Router has now been configured for the InnoDB cluster 'pocCluster'.

The following connection information can be used to connect to the cluster.

Classic MySQL protocol connections to cluster 'pocCluster':

- Read/Write Connections: localhost:6446

- Read/Only Connections: localhost:6447

X protocol connections to cluster 'pocCluster':

- Read/Write Connections: localhost:64460

- Read/Only Connections: localhost:64470

4 ports TCP ont été configurés, dont 2 pour les connexions MySQL traditionnelles:

6446 : lectures / écritures pour le noeud primaire
6447 : lectures seules pour les nœuds secondaires (Round-Robin)
Et le pendant pour les connexions avec le protocole X (64460 & 64470), pour une utilisation NoSQL Document Store de MySQL.

Le fichier de configuration du Router contient quelques informations importantes, tel que le(s) port(s) à utiliser par l’application :

daz@192.168.1.11 ~ $ cat ~/RouterPoC/mysqlrouter.conf
# File automatically generated during MySQL Router bootstrap
[DEFAULT]
logging_folder=/home/daz/RouterPoC/log
runtime_folder=/home/daz/RouterPoC/run
data_folder=/home/daz/RouterPoC/data
keyring_path=/home/daz/RouterPoC/data/keyring
master_key_path=/home/daz/RouterPoC/mysqlrouter.key

[logger]
level = INFO

[metadata_cache:pocCluster]
router_id=3
bootstrap_server_addresses=mysql://192.168.1.22:3306,mysql://192.168.1.50:3306,mysql://192.168.1.48:3306
user=mysql_router3_qeteek4wi41s
metadata_cluster=pocCluster
ttl=300

[routing:pocCluster_default_rw]
bind_address=0.0.0.0
bind_port=6446
destinations=metadata-cache://pocCluster/default?role=PRIMARY
mode=read-write
protocol=classic

[routing:pocCluster_default_ro]
bind_address=0.0.0.0
bind_port=6447
destinations=metadata-cache://pocCluster/default?role=SECONDARY
mode=read-only
protocol=classic

[routing:pocCluster_default_x_rw]
bind_address=0.0.0.0
bind_port=64460
destinations=metadata-cache://pocCluster/default?role=PRIMARY
mode=read-write
protocol=x

[routing:pocCluster_default_x_ro]
bind_address=0.0.0.0
bind_port=64470
destinations=metadata-cache://pocCluster/default?role=SECONDARY
mode=read-only
protocol=x

daz@192.168.1.11 ~ $ cat ~/RouterPoC/mysqlrouter.conf

# File automatically generated during MySQL Router bootstrap

[DEFAULT]

logging_folder=/home/daz/RouterPoC/log

runtime_folder=/home/daz/RouterPoC/run

data_folder=/home/daz/RouterPoC/data

keyring_path=/home/daz/RouterPoC/data/keyring

master_key_path=/home/daz/RouterPoC/mysqlrouter.key

[logger]

level = INFO

[metadata_cache:pocCluster]

router_id=3

bootstrap_server_addresses=mysql://192.168.1.22:3306,mysql://192.168.1.50:3306,mysql://192.168.1.48:3306

user=mysql_router3_qeteek4wi41s

metadata_cluster=pocCluster

ttl=300

[routing:pocCluster_default_rw]

bind_address=0.0.0.0

bind_port=6446

destinations=metadata-cache://pocCluster/default?role=PRIMARY

mode=read-write

protocol=classic

[routing:pocCluster_default_ro]

bind_address=0.0.0.0

bind_port=6447

destinations=metadata-cache://pocCluster/default?role=SECONDARY

mode=read-only

protocol=classic

[routing:pocCluster_default_x_rw]

bind_address=0.0.0.0

bind_port=64460

destinations=metadata-cache://pocCluster/default?role=PRIMARY

mode=read-write

protocol=x

[routing:pocCluster_default_x_ro]

bind_address=0.0.0.0

bind_port=64470

destinations=metadata-cache://pocCluster/default?role=SECONDARY

mode=read-only

protocol=x

Il est évidemment possible de modifier ce fichier. Par exemple, souvent les applications se connectent au port 3306, ça peut donc avoir du sens de modifier le port du noeud primaire en le passant de 6446 à 3306 :

daz@192.168.1.11 ~ $ grep -A6 pocCluster_default_rw ~/RouterPoC/mysqlrouter.conf
[routing:pocCluster_default_rw]
bind_address=0.0.0.0
bind_port=3306
destinations=metadata-cache://pocCluster/default?role=PRIMARY
mode=read-write
protocol=classic

daz@192.168.1.11 ~ $ grep -A6 pocCluster_default_rw ~/RouterPoC/mysqlrouter.conf

[routing:pocCluster_default_rw]

bind_address=0.0.0.0

bind_port=3306

destinations=metadata-cache://pocCluster/default?role=PRIMARY

mode=read-write

protocol=classic

Ton application va donc (continuer à) se connecter au port 3306, sauf que maintenant ce n’est pas 1 instance, mais bel et bien 3 instances qui sont en backend, et ceci en toute transparence.

Ensuite, il faut démarrer MySQL Router avec le script start.sh

daz@192.168.1.11 ~ $ ~/RouterPoC/start.sh

1	daz@192.168.1.11 ~ $ ~/RouterPoC/start.sh

L’arrêt du Router se fait avec le script stop.sh (mais tu l’avais deviné)

daz@192.168.1.11 ~ $ ~/RouterPoC/stop.sh

1	daz@192.168.1.11 ~ $ ~/RouterPoC/stop.sh

Gestion des nœuds

Quelques commandes qui vont te simplifier la vie…

Récupérer les méta-données d’un cluster

Les méta-données du cluster sont stockées sur les membres dans le schéma mysql_innodb_cluster_metadata :

mysql> SHOW SCHEMAS;
+-------------------------------+
| Database                      |
+-------------------------------+
| information_schema            |
| mysql                         |
| mysql_innodb_cluster_metadata |
| performance_schema            |
| sys                           |
+-------------------------------+
5 rows in set (0,00 sec)


mysql> SHOW TABLES IN mysql_innodb_cluster_metadata;
+-----------------------------------------+
| Tables_in_mysql_innodb_cluster_metadata |
+-----------------------------------------+
| clusters                                |
| hosts                                   |
| instances                               |
| replicasets                             |
| routers                                 |
| schema_version                          |
+-----------------------------------------+
6 rows in set (0,01 sec)


mysql> SELECT cluster_name FROM mysql_innodb_cluster_metadata.clusters;
+--------------+
| cluster_name |
+--------------+
| pocCluster   |
+--------------+

mysql> SHOW SCHEMAS;

+-------------------------------+

| Database |

+-------------------------------+

| information_schema |

| mysql |

| mysql_innodb_cluster_metadata |

| performance_schema |

| sys |

+-------------------------------+

5 rows in set (0,00 sec)

mysql> SHOW TABLES IN mysql_innodb_cluster_metadata;

+-----------------------------------------+

| Tables_in_mysql_innodb_cluster_metadata |

+-----------------------------------------+

| clusters |

| hosts |

| instances |

| replicasets |

| routers |

| schema_version |

+-----------------------------------------+

6 rows in set (0,01 sec)

mysql> SELECT cluster_name FROM mysql_innodb_cluster_metadata.clusters;

+--------------+

| cluster_name |

+--------------+

| pocCluster |

+--------------+

Pour récupérer ces informations dans une nouvelle session il faut utiliser la méthode getCluster :

mysql-js> cluster.status()
ReferenceError: cluster is not defined


mysql-js> var cluster = dba.getCluster('pocCluster')


mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

ReferenceError: cluster is not defined

mysql-js> var cluster = dba.getCluster('pocCluster')

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Failover

Le basculement niveau base de données (changement de primaire) est automatiquement géré par les membres du cluster entre eux.

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Crash du noeud primaire (192.168.1.22)…

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.48:3306", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures. 1 member is not active", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "(MISSING)"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.48:3306",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures. 1 member is not active",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "(MISSING)"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Nouveau primaire élu par le groupe : 192.168.1.48.

Et 192.168.1.22 est porté disparu (MIA).

Un rejoinInstance() est nécessaire pour remettre le membre dans le cluster. Il aura un rôle de secondaire (voir paragraphe suivant Remettre un membre dans le groupe).

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.48:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.48:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Remettre un membre dans le groupe

Un membre avec son instance MySQL qui tourne et qui est correctement configuré peut être remis (après un redémarrage de l’instance par exemple) dans le groupe avec la commande rejoinInstance() :

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures. 1 member is not active", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "(MISSING)"
            }
        }
    }
}


mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')
Please provide the password for 'root@192.168.1.50:3306': 
Analyzing the instance replication state...

The instance '192.168.1.50:3306' is valid for the cluster.
The instance is fully recoverable.

{
    "reason": "recoverable", 
    "state": "ok"
}


mysql-js> cluster.rejoinInstance("root@192.168.1.50:3306", {ipWhitelist: "192.168.1.0/16"})
Rejoining the instance to the InnoDB cluster. Depending on the original
problem that made the instance unavailable, the rejoin operation might not be
successful and further manual steps will be needed to fix the underlying
problem.

Please monitor the output of the rejoin operation and take necessary action if
the instance cannot rejoin.

Please provide the password for 'root@192.168.1.50:3306': 
Rejoining instance to the cluster ...

The instance 'root@192.168.1.50:3306' was successfully rejoined on the cluster.

The instance '192.168.1.50:3306' was successfully added to the MySQL Cluster.


mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures. 1 member is not active",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "(MISSING)"

}

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')

Please provide the password for 'root@192.168.1.50:3306':

Analyzing the instance replication state...

The instance '192.168.1.50:3306' is valid for the cluster.

The instance is fully recoverable.

{

"reason": "recoverable",

"state": "ok"

}

mysql-js> cluster.rejoinInstance("root@192.168.1.50:3306", {ipWhitelist: "192.168.1.0/16"})

Rejoining the instance to the InnoDB cluster. Depending on the original

problem that made the instance unavailable, the rejoin operation might not be

successful and further manual steps will be needed to fix the underlying

problem.

Please monitor the output of the rejoin operation and take necessary action if

the instance cannot rejoin.

Please provide the password for 'root@192.168.1.50:3306':

Rejoining instance to the cluster ...

The instance 'root@192.168.1.50:3306' was successfully rejoined on the cluster.

The instance '192.168.1.50:3306' was successfully added to the MySQL Cluster.

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Supprimer une instance du groupe

Sans grande surprise, c’est la commande removeInstance

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}


mysql-js> cluster.removeInstance("root@192.168.1.50:3306")
The instance will be removed from the InnoDB cluster. Depending on the 
instance being the Seed or not, the Metadata session might become invalid. 
If so, please start a new session to the Metadata Storage R/W instance.

The instance 'root@192.168.1.50:3306' was successfully removed from the cluster.


mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

mysql-js> cluster.removeInstance("root@192.168.1.50:3306")

The instance will be removed from the InnoDB cluster. Depending on the

instance being the Seed or not, the Metadata session might become invalid.

If so, please start a new session to the Metadata Storage R/W instance.

The instance 'root@192.168.1.50:3306' was successfully removed from the cluster.

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

L’instance n’est alors plus listée dans les méta-données :

mysql-js> \sql
Switching to SQL mode... Commands end with ;


mysql-sql> SELECT instance_name FROM mysql_innodb_cluster_metadata.instances;
+-------------------+
| instance_name     |
+-------------------+
| 192.168.1.22:3306 |
| 192.168.1.48:3306 |
+-------------------+
2 rows in set (0.00 sec)

mysql-js> \sql

Switching to SQL mode... Commands end with ;

mysql-sql> SELECT instance_name FROM mysql_innodb_cluster_metadata.instances;

+-------------------+

| instance_name |

+-------------------+

| 192.168.1.22:3306 |

| 192.168.1.48:3306 |

+-------------------+

2 rows in set (0.00 sec)

Pour la remettre dans le groupe, il faut donc rejouer le processus de l’ajout d’instance vu plus haut :

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')
Please provide the password for 'root@192.168.1.50:3306': 
Analyzing the instance replication state...

The instance '192.168.1.50:3306' is valid for the cluster.
The instance is fully recoverable.

{
    "reason": "recoverable", 
    "state": "ok"
}


mysql-js> cluster.addInstance("root@192.168.1.50:3306", {ipWhitelist: "192.168.1.0/16"})
A new instance will be added to the InnoDB cluster. Depending on the amount of
data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@192.168.1.50:3306': 
Adding instance to the cluster ...

The instance 'root@192.168.1.50:3306' was successfully added to the cluster.


mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.checkInstanceState('root@192.168.1.50:3306')

Please provide the password for 'root@192.168.1.50:3306':

Analyzing the instance replication state...

The instance '192.168.1.50:3306' is valid for the cluster.

The instance is fully recoverable.

{

"reason": "recoverable",

"state": "ok"

}

mysql-js> cluster.addInstance("root@192.168.1.50:3306", {ipWhitelist: "192.168.1.0/16"})

A new instance will be added to the InnoDB cluster. Depending on the amount of

data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@192.168.1.50:3306':

Adding instance to the cluster ...

The instance 'root@192.168.1.50:3306' was successfully added to the cluster.

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Perte du quorum

Si le cluster perd plus de la moitié de ses membres il se retrouve dans un état assez désagréable, network partitioning, en clair il faut une intervention externe au groupe pour permettre aux membres restant de continuer à faire leur boulot de serveur MySQL.

Dans notre cas, avec 3 instances, il faut en perdre 2 d’un coup :

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Perte des nœuds (crash) 192.168.1.22 & 192.168.1.50… (Mayday, Mayday, Mayday!!!)

mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.22:3306", 
        "status": "NO_QUORUM", 
        "statusText": "Cluster has no quorum as visible from 'localhost:3306' and cannot process write transactions. 2 members are not active", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "UNREACHABLE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "UNREACHABLE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.22:3306",

"status": "NO_QUORUM",

"statusText": "Cluster has no quorum as visible from 'localhost:3306' and cannot process write transactions. 2 members are not active",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "UNREACHABLE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "UNREACHABLE"

}

Le failover automatique ne peut pas s’enclencher, le membre survivant (192.168.1.48) est en lecture seule. Il faut donc intervenir :

$ mysqlsh --uri=root@192.168.1.48
Creating a Session to 'root@192.168.1.48'
...

mysql-js> var cluster = dba.getCluster("pocCluster")
WARNING: The session is on a Read Only instance.
         Write operations on the InnoDB cluster will not be allowed


mysql-js> cluster.forceQuorumUsingPartitionOf('192.168.1.48:3306')
Restoring replicaset 'default' from loss of quorum, by using the partition composed of [192.168.1.48:3306]

Please provide the password for 'root@192.168.1.48:3306': 
Restoring the InnoDB cluster ...

The InnoDB cluster was successfully restored using the partition from the instance 'root@192.168.1.48:3306'.

WARNING: To avoid a split-brain scenario, ensure that all other members of the replicaset are removed or joined back to the group that was restored.


mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.48:3306", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures. 2 members are not active", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "(MISSING)"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "(MISSING)"
            }
        }
    }
}

$ mysqlsh --uri=root@192.168.1.48

Creating a Session to 'root@192.168.1.48'

...

mysql-js> var cluster = dba.getCluster("pocCluster")

WARNING: The session is on a Read Only instance.

Write operations on the InnoDB cluster will not be allowed

mysql-js> cluster.forceQuorumUsingPartitionOf('192.168.1.48:3306')

Restoring replicaset 'default' from loss of quorum, by using the partition composed of [192.168.1.48:3306]

Please provide the password for 'root@192.168.1.48:3306':

Restoring the InnoDB cluster ...

The InnoDB cluster was successfully restored using the partition from the instance 'root@192.168.1.48:3306'.

WARNING: To avoid a split-brain scenario, ensure that all other members of the replicaset are removed or joined back to the group that was restored.

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.48:3306",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures. 2 members are not active",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "(MISSING)"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "(MISSING)"

}

Evidemment, sauf si tu es joueur 🙂 , il faut éviter de rester trop longtemps dans cet état.

Une fois les instances en condition, il faut utiliser rejoinInstance() pour remettre les membres dans le cluster, en tant que secondaire (voir paragraphe Remettre un membre dans le groupe).

Repartir après un arrêt total du cluster

La perte du quorum est une chose, mais il y a pire, perdre tout les nœuds…

En cas d’arrêt total du cluster i.e. toutes les instances sont éteintes, il faut utiliser, une fois les instances MySQL de nouveau démarrées rebootClusterFromCompleteOutage() :

$ mysqlsh --uri=root@192.168.1.48
Creating a Session to 'root@mysql_node1'
...

mysql-js> var cluster = dba.rebootClusterFromCompleteOutage('pocCluster')
Reconfiguring the cluster 'pocCluster' from complete outage...

The instance '192.168.1.22:3306' was part of the cluster configuration.
Would you like to rejoin it to the cluster? [y|N]: y

The instance '192.168.1.50:3306' was part of the cluster configuration.
Would you like to rejoin it to the cluster? [y|N]: y


The cluster was successfully rebooted.


mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "192.168.1.48:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "192.168.1.22:3306": {
                "address": "192.168.1.22:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.48:3306": {
                "address": "192.168.1.48:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "192.168.1.50:3306": {
                "address": "192.168.1.50:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

$ mysqlsh --uri=root@192.168.1.48

Creating a Session to 'root@mysql_node1'

...

mysql-js> var cluster = dba.rebootClusterFromCompleteOutage('pocCluster')

Reconfiguring the cluster 'pocCluster' from complete outage...

The instance '192.168.1.22:3306' was part of the cluster configuration.

Would you like to rejoin it to the cluster? [y|N]: y

The instance '192.168.1.50:3306' was part of the cluster configuration.

Would you like to rejoin it to the cluster? [y|N]: y

The cluster was successfully rebooted.

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "192.168.1.48:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"192.168.1.22:3306": {

"address": "192.168.1.22:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.48:3306": {

"address": "192.168.1.48:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"192.168.1.50:3306": {

"address": "192.168.1.50:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Le membre sur lequel la commande est exécutée est le nouveau primaire.

Dans la même thématique :

Thanks for using MySQL!

Filed under:
Group Replication, Haute Disponibilité, InnoDB Cluster, MySQL by Olivier DASINI

Slides du meetup « Mise en bouche PHP »

mai 3, 2017

Mardi soir, j’ai eu le plaisir d’échanger avec la communauté PHP parisienne et de présenter MySQL InnoDB Cluster lors du meetup « Mise en bouche PHP » organisé par l’AFUP Paris et sponsorisé par Oracle MySQL.

Voici les slides

MySQL InnoDB Cluster – Meetup Oracle MySQL / AFUP Paris from Olivier DASINI

Prochain rendez-vous, le 18 mai pour le PHP Tour 2017 à Nantes.

Thanks for using MySQL!

Filed under:
Conférence, Group Replication, Haute Disponibilité, InnoDB Cluster, PHP, Présentation by Olivier DASINI

Mes confs MySQL pour avril et mai 2017

avril 24, 2017

J’aurai le plaisir de vous rencontrer (et accessoirement de parler de MySQL) les 25 avril, 2 et 18 mai.

MySQL The world's most popular open source database

MySQL Day Paris

Toutes l’info MySQL par l’équipe MySQL France

Info et inscription

Mise en bouche PHP – Paris

Meetup organisé par l’antenne AFUP Paris

Info et inscription

PHP tour 2017 – Nantes

La prochaine étape du cycle de conférences itinérant, réunissant toutes les communautés PHP se déroule à Nantes

Info et inscription

Filed under:
Conférence, InnoDB Cluster, MySQL, PHP by Olivier DASINI

Adopte un… cluster MySQL Group Replication

avril 10, 2017

Tags: Docker, group replication, Haute disponibilité, InnoDB Cluster, router, shell

Autant le dire tout de suite, rien avoir avec un site internet de rencontre en ligne! 🙂
C’est bel et bien un nouvel article dans la série, Haute Disponibilité avec MySQL.

Au menu d’aujourd’hui : comment passer de l’administration « manuelle » de votre solution HA MySQL Group Replication à une administration plus simple, plus fun mais surtout facilement automatisable avec le pack MySQL InnoDB Cluster. En clair, on va voir comment utiliser MySQL Shell pour l’administration et l’orchestration du cluster et MySQL Router pour rediriger automatiquement les transactions de l’application vers le noeud primaire du cluster.

Quelques pré-requis sont nécessaire pour optimiser ta compréhension de cet article, je te conseille donc la lecture préalable des articles suivants:

Note:
L’article traite de MySQL InnoDB Cluster, HA natif de MySQL Server, solution à ne pas confondre avec MySQL NDB Cluster.

Le contexte

Pour ce PoC, j’ai un cluster MySQL Group Replication de 3 nœuds, fonctionnel, en mode « Single Primary » (déployé avec Docker Compose):

Instance 1 : mysql_node1 (172.19.0.2)
Instance 2 : mysql_node2 (172.19.0.4)
Instance 3 : mysql_node3 (172.19.0.3)

$ docker inspect mysql_node1 | grep IPAddress | tail -1
                    "IPAddress": "172.19.0.2",
$ docker inspect mysql_node2 | grep IPAddress | tail -1
                    "IPAddress": "172.19.0.4",
$ docker inspect mysql_node3 | grep IPAddress | tail -1
                    "IPAddress": "172.19.0.3",

$ docker inspect mysql_node1 | grep IPAddress | tail -1

"IPAddress": "172.19.0.2",

$ docker inspect mysql_node2 | grep IPAddress | tail -1

"IPAddress": "172.19.0.4",

$ docker inspect mysql_node3 | grep IPAddress | tail -1

"IPAddress": "172.19.0.3",

MySQL Router et mon application (simulée avec le client texte mysql) sont sur la machine host (par commodité). C’est également le cas de MySQL Shell.

En ce qui concerne les versions des softs:

MySQL Server 5.7.17
MySQL Router 2.1.2 rc
MySQL Shell 1.0.8-rc

Docker 1.12.6 & Docker-compose 1.11.2. Docker est hors du cadre de cet article, mais tu trouveras à la fin de cet article le fichier docker-compose.yml utilisé.

Ah oui, j’ai failli oublier :

TL;DR
Tu as un cluster MySQL Group Replication configuré/administré manuellement et qui tourne. Tu peux l’administrer / le configurer avec MySQL Shell et gérer le routage des requêtes applicatives avec MySQL Router, ces 3 composants forment MySQL InnoDB Cluster.

MySQL InnoDB Cluster Overview

MySQL Group Replication

Les étapes de déploiement du cluster Group Replication ont déjà été traitées ici.

Voici mes 3 instances MySQL 5.7

$ docker ps
CONTAINER ID        IMAGE               COMMAND                  CREATED              STATUS              PORTS                     NAMES
f4c4aa2d3726        mysql:5.7           "docker-entrypoint.sh"   About a minute ago   Up About a minute   0.0.0.0:14002->3306/tcp   mysql_node2
2304f0e44d4c        mysql:5.7           "docker-entrypoint.sh"   About a minute ago   Up About a minute   0.0.0.0:14003->3306/tcp   mysql_node3
fb6ae3c76a06        mysql:5.7           "docker-entrypoint.sh"   About a minute ago   Up About a minute   0.0.0.0:14001->3306/tcp   mysql_node1

$ docker ps

CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES

f4c4aa2d3726 mysql:5.7 "docker-entrypoint.sh" About a minute ago Up About a minute 0.0.0.0:14002->3306/tcp mysql_node2

2304f0e44d4c mysql:5.7 "docker-entrypoint.sh" About a minute ago Up About a minute 0.0.0.0:14003->3306/tcp mysql_node3

fb6ae3c76a06 mysql:5.7 "docker-entrypoint.sh" About a minute ago Up About a minute 0.0.0.0:14001->3306/tcp mysql_node1

MySQL 5.7.17 plus précisément.

 $ docker exec -it mysql_node1 mysql -uroot -p -e"SELECT version();"
Enter password: 
+------------+
| version()  |
+------------+
| 5.7.17-log |
+------------+

$ docker exec -it mysql_node1 mysql -uroot -p -e"SELECT version();"

Enter password:

+------------+

| version() |

+------------+

| 5.7.17-log |

+------------+

A quoi ressemble mon cluster Group Replication ?

Je peux avoir la description de l’architecture avec la table performance_schema.replication_group_members :

docker exec -it mysql_node1 mysql -uroot -p -e"SELECT * FROM performance_schema.replication_group_members\G" 
Enter password: 
*************************** 1. row ***************************
CHANNEL_NAME: group_replication_applier
   MEMBER_ID: d300be14-1797-11e7-a22e-0242ac130002
 MEMBER_HOST: mysql_node1
 MEMBER_PORT: 3306
MEMBER_STATE: ONLINE
*************************** 2. row ***************************
CHANNEL_NAME: group_replication_applier
   MEMBER_ID: d3100569-1797-11e7-a278-0242ac130004
 MEMBER_HOST: mysql_node2
 MEMBER_PORT: 3306
MEMBER_STATE: ONLINE
*************************** 3. row ***************************
CHANNEL_NAME: group_replication_applier
   MEMBER_ID: d321b2db-1797-11e7-a16f-0242ac130003
 MEMBER_HOST: mysql_node3
 MEMBER_PORT: 3306
MEMBER_STATE: ONLINE

docker exec -it mysql_node1 mysql -uroot -p -e"SELECT * FROM performance_schema.replication_group_members\G"

Enter password:

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: d300be14-1797-11e7-a22e-0242ac130002

MEMBER_HOST: mysql_node1

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

*************************** 2. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: d3100569-1797-11e7-a278-0242ac130004

MEMBER_HOST: mysql_node2

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

*************************** 3. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: d321b2db-1797-11e7-a16f-0242ac130003

MEMBER_HOST: mysql_node3

MEMBER_PORT: 3306

MEMBER_STATE: ONLINE

L’identification du noeud primaire peut se faire de la manière suivante :

$ docker exec -it mysql_node1 mysql -uroot -p -e"SELECT MEMBER_ID, MEMBER_HOST, MEMBER_STATE FROM performance_schema.replication_group_members INNER JOIN performance_schema.global_status ON (MEMBER_ID = VARIABLE_VALUE) WHERE VARIABLE_NAME='group_replication_primary_member'\G" 
Enter password: 
*************************** 1. row ***************************
   MEMBER_ID: 8b5bad71-1720-11e7-94f0-0242ac130002
 MEMBER_HOST: mysql_node1
MEMBER_STATE: ONLINE

$ docker exec -it mysql_node1 mysql -uroot -p -e"SELECT MEMBER_ID, MEMBER_HOST, MEMBER_STATE FROM performance_schema.replication_group_members INNER JOIN performance_schema.global_status ON (MEMBER_ID = VARIABLE_VALUE) WHERE VARIABLE_NAME='group_replication_primary_member'\G"

Enter password:

*************************** 1. row ***************************

MEMBER_ID: 8b5bad71-1720-11e7-94f0-0242ac130002

MEMBER_HOST: mysql_node1

MEMBER_STATE: ONLINE

Le noeud mysql_node1 est donc en mode lecture écriture aka le noeud primaire (cette info nous sera utile pour la suite) et les 2 autres en lecture seule (super read only activé):

$ docker exec -it mysql_node2 mysql -uroot -p -e"CREATE SCHEMA gr_test"
Enter password: 
ERROR 1290 (HY000) at line 1: The MySQL server is running with the --super-read-only option so it cannot execute this statement

$ docker exec -it mysql_node2 mysql -uroot -p -e"CREATE SCHEMA gr_test"

Enter password:

ERROR 1290 (HY000) at line 1: The MySQL server is running with the --super-read-only option so it cannot execute this statement

$ docker exec -it mysql_node1 mysql -uroot -p -e"CREATE SCHEMA gr_test;"
Enter password: 

$ docker exec -it mysql_node2 mysql -uroot -p -e"SHOW SCHEMAS;"
Enter password: 
+--------------------+
| Database           |
+--------------------+
| information_schema |
| gr_test            |
| mysql              |
| performance_schema |
| sys                |
+--------------------+

$ docker exec -it mysql_node1 mysql -uroot -p -e"CREATE SCHEMA gr_test;"

Enter password:

$ docker exec -it mysql_node2 mysql -uroot -p -e"SHOW SCHEMAS;"

Enter password:

+--------------------+

| Database |

+--------------------+

| information_schema |

| gr_test |

| mysql |

| performance_schema |

| sys |

+--------------------+

On a donc un cluster MySQL Group Replication avec 3 nœuds online.
Le membre mysql_node1 est (pour le moment) le primaire, mysql_node2 et mysql_node3 sont les secondaires.

L’étape suivant consistera à gérer le cluster avec MySQL Shell.

MySQL Shell, interface pour gérer son cluster

On va se connecter avec le client MySQL Shell au noeud primaire :

$ # Connect to the primary node : mysql_node1
$ mysqlsh --uri=root@mysql_node1
Creating a Session to 'root@mysql_node1'super read only
Enter password: 
Classic Session successfully established. No default schema selected.
Welcome to MySQL Shell 1.0.8-rc

Copyright (c) 2016, 2017, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.
mysql-js>

$ # Connect to the primary node : mysql_node1

$ mysqlsh --uri=root@mysql_node1

Creating a Session to 'root@mysql_node1'super read only

Enter password:

Classic Session successfully established. No default schema selected.

Welcome to MySQL Shell 1.0.8-rc

Oracle is a registered trademark of Oracle Corporation and/or its

affiliates. Other names may be trademarks of their respective

owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.

mysql-js>

Ensuite, je « crée » mon cluster, en fait je vais rendre persistante les informations relatives à l’architecture du groupe dans mon cluster (plus d’info sur ce sujet plus bas).

mysql-js> var cluster=dba.createCluster('pocCluster', {adoptFromGR: true, ipWhitelist:'172.19.0.0/16'})
A new InnoDB cluster will be created on instance 'root@mysql_node1:3306'.

Creating InnoDB cluster 'pocCluster' on 'root@mysql_node1:3306'...
Adding Seed Instance...

Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.
At least 3 instances are needed for the cluster to be able to withstand up to
one server failure.

mysql-js> var cluster=dba.createCluster('pocCluster', {adoptFromGR: true, ipWhitelist:'172.19.0.0/16'})

A new InnoDB cluster will be created on instance 'root@mysql_node1:3306'.

Creating InnoDB cluster 'pocCluster' on 'root@mysql_node1:3306'...

Adding Seed Instance...

Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.

At least 3 instances are needed for the cluster to be able to withstand up to

one server failure.

La méthode createCluster() prends comme paramètres, le nom du cluster (pocCluster) ainsi que des paramètres optionnels comme ipWhitelist (172.19.0.0/16)…

Pour plus de détails connecte toi à MySQL Shell (mysqlsh) et tape : dba.help(‘createCluster’)

$ mysqlsh
*Welcome to MySQL Shell 1.0.8-rc

Copyright (c) 2016, 2017, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.

mysql-js> dba.help('createCluster')

$ mysqlsh

*Welcome to MySQL Shell 1.0.8-rc

Oracle is a registered trademark of Oracle Corporation and/or its

affiliates. Other names may be trademarks of their respective

owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.

mysql-js> dba.help('createCluster')

Vérifions l’état du cluster

mysql-js> /* Check cluster status */
mysql-js> cluster.status()
{
    "clusterName": "pocCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "mysql_node1:3306", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "mysql_node1:3306": {
                "address": "mysql_node1:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "mysql_node2:3306": {
                "address": "mysql_node2:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "mysql_node3:3306": {
                "address": "mysql_node3:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> /* Check cluster status */

mysql-js> cluster.status()

{

"clusterName": "pocCluster",

"defaultReplicaSet": {

"name": "default",

"primary": "mysql_node1:3306",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"mysql_node1:3306": {

"address": "mysql_node1:3306",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"mysql_node2:3306": {

"address": "mysql_node2:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"mysql_node3:3306": {

"address": "mysql_node3:3306",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

MySQL Shell nous confirme ce que nous savions déjà:

je m’auto cite; un grand (1m89) DBA à dit un jour :

« On a donc un cluster MySQL Group Replication déployé avec 3 nœuds online. Le membre mysql_node1 est (pour le moment) le primaire, mysql_node2 et mysql_node3 sont les secondaires. »

En zoomant dans les entrailles du groupe, on constate que la méthode createCluster() a écrit des données dans le cluster :

$ docker exec -it mysql_node1 mysql -uroot -p -e"SHOW SCHEMAS; SHOW TABLES IN mysql_innodb_cluster_metadata;"
Enter password: 
+-------------------------------+
| Database                      |
+-------------------------------+
| information_schema            |
| gr_test                       |
| mysql                         |
| mysql_innodb_cluster_metadata |
| performance_schema            |
| sys                           |
+-------------------------------+

+-----------------------------------------+
| Tables_in_mysql_innodb_cluster_metadata |
+-----------------------------------------+
| clusters                                |
| hosts                                   |
| instances                               |
| replicasets                             |
| routers                                 |
| schema_version                          |
+-----------------------------------------+

$ docker exec -it mysql_node1 mysql -uroot -p -e"SHOW SCHEMAS; SHOW TABLES IN mysql_innodb_cluster_metadata;"

Enter password:

+-------------------------------+

| Database |

+-------------------------------+

| information_schema |

| gr_test |

| mysql |

| mysql_innodb_cluster_metadata |

| performance_schema |

| sys |

+-------------------------------+

+-----------------------------------------+

| Tables_in_mysql_innodb_cluster_metadata |

+-----------------------------------------+

| clusters |

| hosts |

| instances |

| replicasets |

| routers |

| schema_version |

+-----------------------------------------+

Le schéma mysql_innodb_cluster_metadata a donc été créé pour contenir les informations relatives au cluster.

Le nom des tables est assez explicite :

hosts

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.hosts\G"
Enter password: 
*************************** 1. row ***************************
           host_id: 6
         host_name: mysql_node1
        ip_address: 
 public_ip_address: NULL
          location: 
        attributes: NULL
admin_user_account: NULL
*************************** 2. row ***************************
           host_id: 13
         host_name: mysql_node2
        ip_address: 
 public_ip_address: NULL
          location: 
        attributes: NULL
admin_user_account: NULL
*************************** 3. row ***************************
           host_id: 20
         host_name: mysql_node3
        ip_address: 
 public_ip_address: NULL
          location: 
        attributes: NULL
admin_user_account: NULL

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.hosts\G"

Enter password:

*************************** 1. row ***************************

host_id: 6

host_name: mysql_node1

ip_address:

public_ip_address: NULL

location:

attributes: NULL

admin_user_account: NULL

*************************** 2. row ***************************

host_id: 13

host_name: mysql_node2

ip_address:

public_ip_address: NULL

location:

attributes: NULL

admin_user_account: NULL

*************************** 3. row ***************************

host_id: 20

host_name: mysql_node3

ip_address:

public_ip_address: NULL

location:

attributes: NULL

admin_user_account: NULL

clusters

 $ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.clusters\G"
Enter password: 
*************************** 1. row ***************************
         cluster_id: 6
       cluster_name: pocCluster
 default_replicaset: 6
        description: Default Cluster
mysql_user_accounts: NULL
            options: {"adminType": "local"}
         attributes: {"default": true}

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.clusters\G"

Enter password:

*************************** 1. row ***************************

cluster_id: 6

cluster_name: pocCluster

default_replicaset: 6

description: Default Cluster

mysql_user_accounts: NULL

options: {"adminType": "local"}

attributes: {"default": true}

replicasets

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.replicasets\G"
Enter password: 
*************************** 1. row ***************************
  replicaset_id: 6
     cluster_id: 6
replicaset_type: gr
  topology_type: pm
replicaset_name: default
         active: 1
     attributes: {"adopted": "true", "group_replication_group_name": "4e0f05b7-d9d0-11e6-87cf-002710cccc64"}
    description: NULL

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.replicasets\G"

Enter password:

*************************** 1. row ***************************

replicaset_id: 6

cluster_id: 6

replicaset_type: gr

topology_type: pm

replicaset_name: default

active: 1

attributes: {"adopted": "true", "group_replication_group_name": "4e0f05b7-d9d0-11e6-87cf-002710cccc64"}

description: NULL

instances

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.instances\G"
Enter password: 
*************************** 1. row ***************************
      instance_id: 6
          host_id: 6
    replicaset_id: 6
mysql_server_uuid: d300be14-1797-11e7-a22e-0242ac130002
    instance_name: mysql_node1:3306
             role: HA
           weight: NULL
        addresses: {"mysqlX": "mysql_node1:33060", "grLocal": "mysql_node1:4999", "mysqlClassic": "mysql_node1:3306"}
       attributes: NULL
    version_token: NULL
      description: NULL
*************************** 2. row ***************************
      instance_id: 13
          host_id: 13
    replicaset_id: 6
mysql_server_uuid: d3100569-1797-11e7-a278-0242ac130004
    instance_name: mysql_node2:3306
             role: HA
           weight: NULL
        addresses: {"mysqlX": "mysql_node2:33060", "grLocal": "mysql_node2:4999", "mysqlClassic": "mysql_node2:3306"}
       attributes: NULL
    version_token: NULL
      description: NULL
*************************** 3. row ***************************
      instance_id: 20
          host_id: 20
    replicaset_id: 6
mysql_server_uuid: d321b2db-1797-11e7-a16f-0242ac130003
    instance_name: mysql_node3:3306
             role: HA
           weight: NULL
        addresses: {"mysqlX": "mysql_node3:33060", "grLocal": "mysql_node3:4999", "mysqlClassic": "mysql_node3:3306"}
       attributes: NULL
    version_token: NULL
      description: NULL

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.instances\G"

Enter password:

*************************** 1. row ***************************

instance_id: 6

host_id: 6

replicaset_id: 6

mysql_server_uuid: d300be14-1797-11e7-a22e-0242ac130002

instance_name: mysql_node1:3306

role: HA

weight: NULL

addresses: {"mysqlX": "mysql_node1:33060", "grLocal": "mysql_node1:4999", "mysqlClassic": "mysql_node1:3306"}

attributes: NULL

version_token: NULL

description: NULL

*************************** 2. row ***************************

instance_id: 13

host_id: 13

replicaset_id: 6

mysql_server_uuid: d3100569-1797-11e7-a278-0242ac130004

instance_name: mysql_node2:3306

role: HA

weight: NULL

addresses: {"mysqlX": "mysql_node2:33060", "grLocal": "mysql_node2:4999", "mysqlClassic": "mysql_node2:3306"}

attributes: NULL

version_token: NULL

description: NULL

*************************** 3. row ***************************

instance_id: 20

host_id: 20

replicaset_id: 6

mysql_server_uuid: d321b2db-1797-11e7-a16f-0242ac130003

instance_name: mysql_node3:3306

role: HA

weight: NULL

addresses: {"mysqlX": "mysql_node3:33060", "grLocal": "mysql_node3:4999", "mysqlClassic": "mysql_node3:3306"}

attributes: NULL

version_token: NULL

description: NULL

Déploiement de MySQL Router

Le déploiement du router est trivial, il faut pour commencer le bootstrapper au cluster, c’est à dire le lier au cluster en le connectant aux méta-données :

$ mysqlrouter --bootstrap root@mysql_node1:3306 --directory routerDocker --name routerDocker
[sudo] password for daz: 
Please enter MySQL password for root: 

Bootstrapping MySQL Router instance at /home/daz/routerDocker...
MySQL Router 'routerDocker' has now been configured for the InnoDB cluster 'pocCluster'.

The following connection information can be used to connect to the cluster.

Classic MySQL protocol connections to cluster 'pocCluster':
- Read/Write Connections: localhost:6446
- Read/Only Connections: localhost:6447

X protocol connections to cluster 'pocCluster':
- Read/Write Connections: localhost:64460
- Read/Only Connections: localhost:64470

$ mysqlrouter --bootstrap root@mysql_node1:3306 --directory routerDocker --name routerDocker

[sudo] password for daz:

Please enter MySQL password for root:

Bootstrapping MySQL Router instance at /home/daz/routerDocker...

MySQL Router 'routerDocker' has now been configured for the InnoDB cluster 'pocCluster'.

The following connection information can be used to connect to the cluster.

Classic MySQL protocol connections to cluster 'pocCluster':

- Read/Write Connections: localhost:6446

- Read/Only Connections: localhost:6447

X protocol connections to cluster 'pocCluster':

- Read/Write Connections: localhost:64460

- Read/Only Connections: localhost:64470

Les paramètres directory et name sont optionnels.

Lancer MySQL Router :

$ mysqlrouter -c ~/routerDocker/mysqlrouter.conf &

1	$ mysqlrouter -c ~/routerDocker/mysqlrouter.conf &

L’application doit se connecter (par défaut) au port 6446 (écritures et lectures vers le noeud primaire). En cas de besoin de read scalability, les lectures peuvent être dirigées vers le port 6447.

Inspectons de nouveau les méta données :

routers

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.routers\G"
Enter password: 
*************************** 1. row ***************************
  router_id: 6
router_name: routerDocker
    host_id: 27
 attributes: NULL

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.routers\G"

Enter password:

*************************** 1. row ***************************

router_id: 6

router_name: routerDocker

host_id: 27

attributes: NULL

hosts

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.hosts\G"
Enter password: 
*************************** 1. row ***************************
           host_id: 6
         host_name: mysql_node1
        ip_address: 
 public_ip_address: NULL
          location: 
        attributes: NULL
admin_user_account: NULL
*************************** 2. row ***************************
           host_id: 13
         host_name: mysql_node2
        ip_address: 
 public_ip_address: NULL
          location: 
        attributes: NULL
admin_user_account: NULL
*************************** 3. row ***************************
           host_id: 20
         host_name: mysql_node3
        ip_address: 
 public_ip_address: NULL
          location: 
        attributes: NULL
admin_user_account: NULL
*************************** 4. row ***************************
           host_id: 27
         host_name: 
        ip_address: NULL
 public_ip_address: NULL
          location: 
        attributes: {"registeredFrom": "mysql-router"}
admin_user_account: NULL

$ docker exec -it mysql_node1 mysql -uroot -p -e" SELECT * FROM mysql_innodb_cluster_metadata.hosts\G"

Enter password:

*************************** 1. row ***************************

host_id: 6

host_name: mysql_node1

ip_address:

public_ip_address: NULL

location:

attributes: NULL

admin_user_account: NULL

*************************** 2. row ***************************

host_id: 13

host_name: mysql_node2

ip_address:

public_ip_address: NULL

location:

attributes: NULL

admin_user_account: NULL

*************************** 3. row ***************************

host_id: 20

host_name: mysql_node3

ip_address:

public_ip_address: NULL

location:

attributes: NULL

admin_user_account: NULL

*************************** 4. row ***************************

host_id: 27

host_name:

ip_address: NULL

public_ip_address: NULL

location:

attributes: {"registeredFrom": "mysql-router"}

admin_user_account: NULL

Voilà, mon cluster Group Replication paramétré « à la main » fait maintenant partie intégrante de mon InnoDB Cluster, je peux donc l’administrer avec MySQL Shell et je peux vous assurer que c’est vraiment pratique.

Mais ceci est une autre histoire et fera l’objet d’un autre article 🙂

Annexe

Le fichier docker-compose est le suivant :

version: '2'
services:
  node1:
    container_name: mysql_node1
    image: "mysql:5.7"
    volumes:
      - ~/Documents/Docker/confdir/mysql1:/etc/mysql/conf.d
    ports:
      - "14001:3306"
    environment:
      - MYSQL_ROOT_PASSWORD=root
    networks:
      app_net:
        ipv4_address: 172.19.0.2
  node2:
    container_name: mysql_node2
    image: "mysql:5.7"
    volumes:
      - ~/Documents/Docker/confdir/mysql2:/etc/mysql/conf.d
    ports:
      - "14002:3306"
    environment:
      - MYSQL_ROOT_PASSWORD=root
    networks:
      app_net:
        ipv4_address: 172.19.0.4
  node3:
    container_name: mysql_node3
    image: "mysql:5.7"
    volumes:
      - ~/Documents/Docker/confdir/mysql3:/etc/mysql/conf.d
    ports:
      - "14003:3306"
    environment:
      - MYSQL_ROOT_PASSWORD=root
    networks:
      app_net:
        ipv4_address: 172.19.0.3

networks:
  app_net:
    driver: bridge
    ipam:
      driver: default
      config:
      -
        subnet: 172.19.0.0/24

version: '2'

services:

node1:

container_name: mysql_node1

image: "mysql:5.7"

volumes:

- ~/Documents/Docker/confdir/mysql1:/etc/mysql/conf.d

ports:

- "14001:3306"

environment:

- MYSQL_ROOT_PASSWORD=root

networks:

app_net:

ipv4_address: 172.19.0.2

node2:

container_name: mysql_node2

image: "mysql:5.7"

volumes:

- ~/Documents/Docker/confdir/mysql2:/etc/mysql/conf.d

ports:

- "14002:3306"

environment:

- MYSQL_ROOT_PASSWORD=root

networks:

app_net:

ipv4_address: 172.19.0.4

node3:

container_name: mysql_node3

image: "mysql:5.7"

volumes:

- ~/Documents/Docker/confdir/mysql3:/etc/mysql/conf.d

ports:

- "14003:3306"

environment:

- MYSQL_ROOT_PASSWORD=root

networks:

app_net:

ipv4_address: 172.19.0.3

networks:

app_net:

driver: bridge

ipam:

driver: default

config:

subnet: 172.19.0.0/24

Thanks for using MySQL!

Filed under:
Group Replication, Haute Disponibilité, InnoDB Cluster, MySQL by Olivier DASINI

Tester MySQL InnoDB Cluster

mars 13, 2017

Tags: group replication, Haute disponibilité, InnoDB Cluster, router, shell

MySQL InnoDB Cluster est la (future) solution out-of-the-box HA de MySQL (à ne pas confondre avec MySQL NDB Cluster). Ce produit est composé de 3 éléments :

MySQL Group Replication
- Plugin de réplication multi-maître, avec résolution de conflits et basculement (failover) automatique.
MySQL Router
- Middleware légé et performant qui fournit un routage transparent entre l’application et le cluster.
MySQL Shell
- Client interactif Javascript, Python et SQL qui permet d’administrer le cluster.

MySQL InnoDB Cluster Architecture
MySQL Group Replication est GA et peut donc être utilisé tel quel hors MySQL InnoDB Cluster (voir Déployer un cluster MySQL Group Replication).

Ce n’est par contre pas encore le cas pour les 2 autres composants, MySQL Shell et MySQL Router qui sont en Release Candidate (RC), il n’est donc pas recommandé à ce jour de les utiliser dans un environnement de production.

Note:
L’article traite de MySQL InnoDB Cluster, HA natif de MySQL Server, solution à ne pas confondre avec MySQL NDB Cluster.

Installer MySQL InnoDB Cluster

Dans le cadre de cet article, les versions utilisées sont:

MySQL Server : 5.7.17
MySQL Shell : 1.0.8-rc
MySQL Router : 2.1.2 rc

Pour utiliser MySQL InnoDB Cluster, il faut simplement installer ces 3 composants :

Déployer les instances de test

MySQL Shell permet de déployer simplement des instances MySQL de test (sandbox).

Connexion avec MySQL Shell :

$ mysqlsh
Welcome to MySQL Shell 1.0.8-rc

Copyright (c) 2016, 2017, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective
owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.
mysql-js>

$ mysqlsh

Welcome to MySQL Shell 1.0.8-rc

Oracle is a registered trademark of Oracle Corporation and/or its

affiliates. Other names may be trademarks of their respective

owners.

Type '\help', '\h' or '\?' for help, type '\quit' or '\q' to exit.

Currently in JavaScript mode. Use \sql to switch to SQL mode and execute queries.

mysql-js>

Déployer la 1ère instance MySQL qui fera partie de notre cluster :

Host : localhost
Port : 3310

mysql-js> dba.deploySandboxInstance(3310)
A new MySQL sandbox instance will be created on this host in 
/home/daz/mysql-sandboxes/3310

Please enter a MySQL root password for the new instance: 
Deploying new MySQL instance...

Instance localhost:3310 successfully deployed and started.
Use shell.connect('root@localhost:3310'); to connect to the instance.

mysql-js> dba.deploySandboxInstance(3310)

A new MySQL sandbox instance will be created on this host in

/home/daz/mysql-sandboxes/3310

Please enter a MySQL root password for the new instance:

Deploying new MySQL instance...

Instance localhost:3310 successfully deployed and started.

Use shell.connect('root@localhost:3310'); to connect to the instance.

Il suffit de rentrer le mot de passe root, puis l’instance est crée dans ~/mysql-sandboxes :

$ ls ~/mysql-sandboxes/3310/
3310.pid my.cnf mysqld mysqld.sock mysqld.sock.lock mysql-files mysqlx.sock mysqlx.sock.lock sandboxdata start.sh stop.sh

1 2	$ ls ~/mysql-sandboxes/3310/ 3310.pid my.cnf mysqld mysqld.sock mysqld.sock.lock mysql-files mysqlx.sock mysqlx.sock.lock sandboxdata start.sh stop.sh

Créons 2 autres instances pour le cluster:

Host : localhost
Ports : 3320 & 3330

mysql-js> dba.deploySandboxInstance(3320)
A new MySQL sandbox instance will be created on this host in 
/home/daz/mysql-sandboxes/3320

Please enter a MySQL root password for the new instance: 
Deploying new MySQL instance...

Instance localhost:3320 successfully deployed and started.
Use shell.connect('root@localhost:3320'); to connect to the instance.

mysql-js> dba.deploySandboxInstance(3330)
A new MySQL sandbox instance will be created on this host in 
/home/daz/mysql-sandboxes/3330

Please enter a MySQL root password for the new instance: 
Deploying new MySQL instance...

Instance localhost:3330 successfully deployed and started.
Use shell.connect('root@localhost:3330'); to connect to the instance.

mysql-js> dba.deploySandboxInstance(3320)

A new MySQL sandbox instance will be created on this host in

/home/daz/mysql-sandboxes/3320

Please enter a MySQL root password for the new instance:

Deploying new MySQL instance...

Instance localhost:3320 successfully deployed and started.

Use shell.connect('root@localhost:3320'); to connect to the instance.

mysql-js> dba.deploySandboxInstance(3330)

A new MySQL sandbox instance will be created on this host in

/home/daz/mysql-sandboxes/3330

Please enter a MySQL root password for the new instance:

Deploying new MySQL instance...

Instance localhost:3330 successfully deployed and started.

Use shell.connect('root@localhost:3330'); to connect to the instance.

On a donc 3 instances MySQL dans notre sandbox.

A Noter que si vous avez déjà un cluster MySQL Group Replication actif, MySQL InnoDB Cluster est capable de l’adopter. Ceci fera l’objet d’un prochain article.

Gérer les instances

D’autres méthodes existent pour gérer les instances:

Stop
- dba.stopSandboxInstance()
Start
- dba.startSandboxInstance()
Kill : permet de simuler le crash d’un nœud
- dba.killSandboxInstance()
Delete : suppression totale de l’instance de la sandbox
- dba.deleteSandboxInstance()

Exemple – Arrêt et suppression d’une instance

mysql-js> dba.stopSandboxInstance(3320)
The MySQL sandbox instance on this host in 
/.../mysql-sandboxes/3320 will be stopped

Please enter the MySQL root password for the instance 'localhost:3320':

Stopping MySQL instance...

Instance localhost:3320 successfully stopped.

mysql-js> dba.deleteSandboxInstance(3320);
The MySQL sandbox instance on this host in 
/.../mysql-sandboxes/3320 will be deleted


Deleting MySQL instance...

Instance localhost:3320 successfully deleted.

mysql-js> dba.stopSandboxInstance(3320)

The MySQL sandbox instance on this host in

/.../mysql-sandboxes/3320 will be stopped

Please enter the MySQL root password for the instance 'localhost:3320':

Stopping MySQL instance...

Instance localhost:3320 successfully stopped.

mysql-js> dba.deleteSandboxInstance(3320);

The MySQL sandbox instance on this host in

/.../mysql-sandboxes/3320 will be deleted

Deleting MySQL instance...

Instance localhost:3320 successfully deleted.

L’aide est disponible dans MySQL Shell avec dba.help()

mysql-js> dba.help()

The global variable 'dba' is used to access the MySQL AdminAPI functionality
and perform DBA operations. It is used for managing MySQL InnoDB clusters.

The following properties are currently supported.

- verbose Enables verbose mode on the Dba operations.


The following functions are currently supported.

- checkInstanceConfiguration
- configureLocalInstance Validates and configures an instance for
cluster usage.
- createCluster Creates a MySQL InnoDB cluster.
- deleteSandboxInstance Deletes an existing MySQL Server instance on
localhost.
- deploySandboxInstance Creates a new MySQL Server instance on
localhost.
- dropMetadataSchema Drops the Metadata Schema.
- getCluster Retrieves a cluster from the Metadata Store.
- help Provides help about this class and it's
members
- killSandboxInstance Kills a running MySQL Server instance on
localhost.
- rebootClusterFromCompleteOutage Reboots a cluster from complete outage.
- resetSession Sets the session object to be used on the
Dba operations.
- startSandboxInstance Starts an existing MySQL Server instance on
localhost.
- stopSandboxInstance Stops a running MySQL Server instance on
localhost.

For more help on a specific function use: dba.help('<functionName>')

e.g. dba.help('deploySandboxInstance')

mysql-js> dba.help()

The global variable 'dba' is used to access the MySQL AdminAPI functionality

and perform DBA operations. It is used for managing MySQL InnoDB clusters.

The following properties are currently supported.

- verbose Enables verbose mode on the Dba operations.

The following functions are currently supported.

- checkInstanceConfiguration

- configureLocalInstance Validates and configures an instance for

cluster usage.

- createCluster Creates a MySQL InnoDB cluster.

- deleteSandboxInstance Deletes an existing MySQL Server instance on

localhost.

- deploySandboxInstance Creates a new MySQL Server instance on

localhost.

- dropMetadataSchema Drops the Metadata Schema.

- getCluster Retrieves a cluster from the Metadata Store.

- help Provides help about this class and it's

members

- killSandboxInstance Kills a running MySQL Server instance on

localhost.

- rebootClusterFromCompleteOutage Reboots a cluster from complete outage.

- resetSession Sets the session object to be used on the

Dba operations.

- startSandboxInstance Starts an existing MySQL Server instance on

localhost.

- stopSandboxInstance Stops a running MySQL Server instance on

localhost.

For more help on a specific function use: dba.help('<functionName>')

e.g. dba.help('deploySandboxInstance')

Vérifier la configuration des instances

Un moyen simple de savoir si les instances ont la configuration requise pour faire partie du cluster est d’utiliser : dba.checkInstanceConfiguration()

mysql-js> dba.checkInstanceConfiguration('root@localhost:3310')
Please provide the password for 'root@localhost:3310': 
Validating instance...

The instance 'localhost:3310' is valid for Cluster usage
{
"status": "ok"
}
mysql-js> dba.checkInstanceConfiguration('root@localhost:3320')
Please provide the password for 'root@localhost:3320': 
Validating instance...

The instance 'localhost:3320' is valid for Cluster usage
{
"status": "ok"
}
mysql-js> dba.checkInstanceConfiguration('root@localhost:3330')
Please provide the password for 'root@localhost:3330': 
Validating instance...

The instance 'localhost:3330' is valid for Cluster usage
{
"status": "ok"
}

mysql-js> dba.checkInstanceConfiguration('root@localhost:3310')

Please provide the password for 'root@localhost:3310':

Validating instance...

The instance 'localhost:3310' is valid for Cluster usage

{

"status": "ok"

}

mysql-js> dba.checkInstanceConfiguration('root@localhost:3320')

Please provide the password for 'root@localhost:3320':

Validating instance...

The instance 'localhost:3320' is valid for Cluster usage

{

"status": "ok"

}

mysql-js> dba.checkInstanceConfiguration('root@localhost:3330')

Please provide the password for 'root@localhost:3330':

Validating instance...

The instance 'localhost:3330' is valid for Cluster usage

{

"status": "ok"

}

Créer le cluster

On a donc 3 instances MySQL, en standalone, configurées et prêtes à se transformer en une base de données distribuée.

Je vais donc me connecter à une de mes instances :

User : root
Host : localhost
Ports : 3310

mysql-js> \c root@localhost:3310
Creating a Session to 'root@localhost:3310'
Enter password: 
Classic Session successfully established. No default schema selected.

mysql-js> \c root@localhost:3310

Creating a Session to 'root@localhost:3310'

Enter password:

Classic Session successfully established. No default schema selected.

Puis commencer la création effective de mon instance MySQL InnoDB Cluster, nommée testcluster

mysql-js> var cluster=dba.createCluster('testcluster')
A new InnoDB cluster will be created on instance 'root@localhost:3310'.

Creating InnoDB cluster 'testcluster' on 'root@localhost:3310'...
Adding Seed Instance...

Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.
At least 3 instances are needed for the cluster to be able to withstand up to
one server failure.

mysql-js> var cluster=dba.createCluster('testcluster')

A new InnoDB cluster will be created on instance 'root@localhost:3310'.

Creating InnoDB cluster 'testcluster' on 'root@localhost:3310'...

Adding Seed Instance...

Cluster successfully created. Use Cluster.addInstance() to add MySQL instances.

At least 3 instances are needed for the cluster to be able to withstand up to

one server failure.

Je me retrouve pour le moment avec un cluster d’1 nœud. Certes, pas encore hautement disponible, mais c’est un début 🙂

La méthode status() me le confirme:

mysql-js> cluster.status()
{
    "clusterName": "testcluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "localhost:3310", 
        "status": "OK_NO_TOLERANCE", 
        "statusText": "Cluster is NOT tolerant to any failures.", 
        "topology": {
            "localhost:3310": {
                "address": "localhost:3310", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "testcluster",

"defaultReplicaSet": {

"name": "default",

"primary": "localhost:3310",

"status": "OK_NO_TOLERANCE",

"statusText": "Cluster is NOT tolerant to any failures.",

"topology": {

"localhost:3310": {

"address": "localhost:3310",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

Avant de lui rajouter des petits copains, on va vérifier que toutes les instances ont la même liste de transactions exécutées:

mysql-js> cluster.checkInstanceState('root@localhost:3310')
Please provide the password for 'root@localhost:3310': 
Analyzing the instance replication state...

The instance 'localhost:3310' is valid for the cluster.
The instance is fully recoverable.

{
"reason": "recoverable", 
"state": "ok"
}

mysql-js> cluster.checkInstanceState('root@localhost:3320')
Please provide the password for 'root@localhost:3320': 
Analyzing the instance replication state...

The instance 'localhost:3320' is valid for the cluster.
The instance is new to Group Replication.

{
"reason": "new", 
"state": "ok"
}

mysql-js> cluster.checkInstanceState('root@localhost:3330')
Please provide the password for 'root@localhost:3330': 
Analyzing the instance replication state...

The instance 'localhost:3330' is valid for the cluster.
The instance is new to Group Replication.

{
"reason": "new", 
"state": "ok"
}

mysql-js> cluster.checkInstanceState('root@localhost:3310')

Please provide the password for 'root@localhost:3310':

Analyzing the instance replication state...

The instance 'localhost:3310' is valid for the cluster.

The instance is fully recoverable.

{

"reason": "recoverable",

"state": "ok"

}

mysql-js> cluster.checkInstanceState('root@localhost:3320')

Please provide the password for 'root@localhost:3320':

Analyzing the instance replication state...

The instance 'localhost:3320' is valid for the cluster.

The instance is new to Group Replication.

{

"reason": "new",

"state": "ok"

}

mysql-js> cluster.checkInstanceState('root@localhost:3330')

Please provide the password for 'root@localhost:3330':

Analyzing the instance replication state...

The instance 'localhost:3330' is valid for the cluster.

The instance is new to Group Replication.

{

"reason": "new",

"state": "ok"

}

Parfait !

Ajouter les autres nœuds

addInstance(), la bien nommée :

Ajout de localhost:3320

mysql-js> cluster.addInstance('root@localhost:3320')
A new instance will be added to the InnoDB cluster. Depending on the amount of
data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@localhost:3320': 
Adding instance to the cluster ...

The instance 'root@localhost:3320' was successfully added to the cluster.

mysql-js> cluster.addInstance('root@localhost:3320')

A new instance will be added to the InnoDB cluster. Depending on the amount of

data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@localhost:3320':

Adding instance to the cluster ...

The instance 'root@localhost:3320' was successfully added to the cluster.

Ajout de localhost:3330

mysql-js> cluster.addInstance('root@localhost:3330')
A new instance will be added to the InnoDB cluster. Depending on the amount of
data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@localhost:3330': 
Adding instance to the cluster ...

The instance 'root@localhost:3330' was successfully added to the cluster.

mysql-js> cluster.addInstance('root@localhost:3330')

A new instance will be added to the InnoDB cluster. Depending on the amount of

data on the cluster this might take from a few seconds to several hours.

Please provide the password for 'root@localhost:3330':

Adding instance to the cluster ...

The instance 'root@localhost:3330' was successfully added to the cluster.

L’architecture de notre cluster est maintenant:

mysql-js> cluster.status()
{
    "clusterName": "testcluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "localhost:3310", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "localhost:3310": {
                "address": "localhost:3310", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "localhost:3320": {
                "address": "localhost:3320", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "localhost:3330": {
                "address": "localhost:3330", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }
}

mysql-js> cluster.status()

{

"clusterName": "testcluster",

"defaultReplicaSet": {

"name": "default",

"primary": "localhost:3310",

"status": "OK",

"statusText": "Cluster is ONLINE and can tolerate up to ONE failure.",

"topology": {

"localhost:3310": {

"address": "localhost:3310",

"mode": "R/W",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"localhost:3320": {

"address": "localhost:3320",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

"localhost:3330": {

"address": "localhost:3330",

"mode": "R/O",

"readReplicas": {},

"role": "HA",

"status": "ONLINE"

}

localhost:3310 est le primaire il est donc le seul à accepter les écritures. Les 2 autres membres ne sont accessibles qu’en lecture.

C’est le comportement par défaut de MySQL Group Replication est donc de MySQL InnoDB Cluster. Pour avoir un cluster en multi-master, il faut le préciser lors de la création du cluster (dba.createCluster()).

Les informations révélées par les différentes commandes exécutée jusqu’ici, sont persistante. Elles sont en stockées dans les nœuds du cluster, dans le schéma mysql_innodb_cluster_metadata :

$ mysql --protocol=tcp -uroot -p -P3310

mysql> show schemas;
+-------------------------------+
| Database                      |
+-------------------------------+
| information_schema            |
| mysql                         |
| mysql_innodb_cluster_metadata |
| performance_schema            |
| sys                           |
+-------------------------------+
5 rows in set (0,00 sec)


mysql> show tables in mysql_innodb_cluster_metadata;
+-----------------------------------------+
| Tables_in_mysql_innodb_cluster_metadata |
+-----------------------------------------+
| clusters                                |
| hosts                                   |
| instances                               |
| replicasets                             |
| routers                                 |
| schema_version                          |
+-----------------------------------------+
6 rows in set (0,00 sec)

$ mysql --protocol=tcp -uroot -p -P3310

mysql> show schemas;

+-------------------------------+

| Database |

+-------------------------------+

| information_schema |

| mysql |

| mysql_innodb_cluster_metadata |

| performance_schema |

| sys |

+-------------------------------+

5 rows in set (0,00 sec)

mysql> show tables in mysql_innodb_cluster_metadata;

+-----------------------------------------+

| Tables_in_mysql_innodb_cluster_metadata |

+-----------------------------------------+

| clusters |

| hosts |

| instances |

| replicasets |

| routers |

| schema_version |

+-----------------------------------------+

6 rows in set (0,00 sec)

Déployer MySQL Router

MySQL Router étant déjà installé, on va le configurer pour l’interfacer avec notre cluster:

$ mysqlrouter --bootstrap root@localhost:3310 --directory routerSandbox
Please enter MySQL password for root:

Bootstrapping system MySQL Router instance...
MySQL Router has now been configured for the InnoDB cluster 'testcluster'.

The following connection information can be used to connect to the cluster.

Classic MySQL protocol connections to cluster 'testcluster':
- Read/Write Connections: localhost:6446
- Read/Only Connections: localhost:6447

X protocol connections to cluster 'testcluster':
- Read/Write Connections: localhost:64460
- Read/Only Connections: localhost:64470

$ mysqlrouter --bootstrap root@localhost:3310 --directory routerSandbox

Please enter MySQL password for root:

Bootstrapping system MySQL Router instance...

MySQL Router has now been configured for the InnoDB cluster 'testcluster'.

The following connection information can be used to connect to the cluster.

Classic MySQL protocol connections to cluster 'testcluster':

- Read/Write Connections: localhost:6446

- Read/Only Connections: localhost:6447

X protocol connections to cluster 'testcluster':

- Read/Write Connections: localhost:64460

- Read/Only Connections: localhost:64470

4 ports TCP ont été configurés, 2 pour les connexions MySQL traditionnelles:

6446 : lectures / écritures pour le noeud primaire
6447 : lectures seules pour les nœuds secondaires (Round-Robin)

Et le pendant pour les connexions avec le protocole X (64460 & 64470), pour une utilisation NoSQL Document Store de MySQL.

Le bootstrap à généré un fichier de configuration pour MySQL Router

$ view ~/routerSandbox/mysqlrouter.conf

# File automatically generated during MySQL Router bootstrap
[DEFAULT]
name=routerSandbox
logging_folder=/home/xxxx/routerSandbox/log
runtime_folder=/home/xxxx/routerSandbox/run
data_folder=/home/xxxx/routerSandbox/data
keyring_path=/home/xxxx/routerSandbox/data/keyring
master_key_path=/home/xxxx/routerSandbox/mysqlrouter.key

[logger]
level = INFO

[metadata_cache:testcluster]
router_id=6
bootstrap_server_addresses=mysql://localhost:3310,mysql://localhost:3320,mysql://localhost:3330
user=mysql_router6_qy5a3dmn5y68
metadata_cluster=testcluster
ttl=300

[routing:testcluster_default_rw]
bind_address=0.0.0.0
bind_port=6446
destinations=metadata-cache://testcluster/default?role=PRIMARY
mode=read-write
protocol=classic

[routing:testcluster_default_ro]
bind_address=0.0.0.0
bind_port=6447
destinations=metadata-cache://testcluster/default?role=SECONDARY
mode=read-only
protocol=classic
...

$ view ~/routerSandbox/mysqlrouter.conf

# File automatically generated during MySQL Router bootstrap

[DEFAULT]

name=routerSandbox

logging_folder=/home/xxxx/routerSandbox/log

runtime_folder=/home/xxxx/routerSandbox/run

data_folder=/home/xxxx/routerSandbox/data

keyring_path=/home/xxxx/routerSandbox/data/keyring

master_key_path=/home/xxxx/routerSandbox/mysqlrouter.key

[logger]

level = INFO

[metadata_cache:testcluster]

router_id=6

bootstrap_server_addresses=mysql://localhost:3310,mysql://localhost:3320,mysql://localhost:3330

user=mysql_router6_qy5a3dmn5y68

metadata_cluster=testcluster

ttl=300

[routing:testcluster_default_rw]

bind_address=0.0.0.0

bind_port=6446

destinations=metadata-cache://testcluster/default?role=PRIMARY

mode=read-write

protocol=classic

[routing:testcluster_default_ro]

bind_address=0.0.0.0

bind_port=6447

destinations=metadata-cache://testcluster/default?role=SECONDARY

mode=read-only

protocol=classic

...

Utiliser MySQL Router

Evidemment il faut le démarrer

$ ~/routerSandbox/start.sh

1	$ ~/routerSandbox/start.sh

Un petit coup d’œil dans les logs:

$ tail -f ~/routerSandbox/log/mysqlrouter.log
 
2017-03-10 19:03:36 INFO    [7f81e420e700] Starting Metadata Cache
2017-03-10 19:03:36 INFO    [7f81e420e700] Connections using ssl_mode 'PREFERRED'
2017-03-10 19:03:36 INFO    [7f81e3a0d700] [routing:testcluster_default_ro] started: listening on 0.0.0.0:6447; read-only
2017-03-10 19:03:36 INFO    [7f81e320c700] [routing:testcluster_default_rw] started: listening on 0.0.0.0:6446; read-write
2017-03-10 19:03:36 INFO    [7f81e2a0b700] [routing:testcluster_default_x_ro] started: listening on 0.0.0.0:64470; read-only
2017-03-10 19:03:36 INFO    [7f81e220a700] [routing:testcluster_default_x_rw] started: listening on 0.0.0.0:64460; read-write
2017-03-10 19:03:36 INFO    [7f81e420e700] Connected with metadata server running on 127.0.0.1:3310
2017-03-10 19:03:36 INFO    [7f81e420e700] Changes detected in cluster 'testcluster' after metadata refresh
2017-03-10 19:03:36 INFO    [7f81e420e700] Metadata for cluster 'testcluster' has 1 replicasets:
2017-03-10 19:03:36 INFO    [7f81e420e700] 'default' (3 members, single-master)
2017-03-10 19:03:36 INFO    [7f81e420e700]     localhost:3310 / 33100 - role=HA mode=RW
2017-03-10 19:03:36 INFO    [7f81e420e700]     localhost:3320 / 33200 - role=HA mode=RO
2017-03-10 19:03:36 INFO    [7f81e420e700]     localhost:3330 / 33300 - role=HA mode=RO
2017-03-10 19:03:36 INFO    [7f81c37fe700] Connected with metadata server running on 127.0.0.1:3310

$ tail -f ~/routerSandbox/log/mysqlrouter.log

2017-03-10 19:03:36 INFO [7f81e420e700] Starting Metadata Cache

2017-03-10 19:03:36 INFO [7f81e420e700] Connections using ssl_mode 'PREFERRED'

2017-03-10 19:03:36 INFO [7f81e3a0d700] [routing:testcluster_default_ro] started: listening on 0.0.0.0:6447; read-only

2017-03-10 19:03:36 INFO [7f81e320c700] [routing:testcluster_default_rw] started: listening on 0.0.0.0:6446; read-write

2017-03-10 19:03:36 INFO [7f81e2a0b700] [routing:testcluster_default_x_ro] started: listening on 0.0.0.0:64470; read-only

2017-03-10 19:03:36 INFO [7f81e220a700] [routing:testcluster_default_x_rw] started: listening on 0.0.0.0:64460; read-write

2017-03-10 19:03:36 INFO [7f81e420e700] Connected with metadata server running on 127.0.0.1:3310

2017-03-10 19:03:36 INFO [7f81e420e700] Changes detected in cluster 'testcluster' after metadata refresh

2017-03-10 19:03:36 INFO [7f81e420e700] Metadata for cluster 'testcluster' has 1 replicasets:

2017-03-10 19:03:36 INFO [7f81e420e700] 'default' (3 members, single-master)

2017-03-10 19:03:36 INFO [7f81e420e700] localhost:3310 / 33100 - role=HA mode=RW

2017-03-10 19:03:36 INFO [7f81e420e700] localhost:3320 / 33200 - role=HA mode=RO

2017-03-10 19:03:36 INFO [7f81e420e700] localhost:3330 / 33300 - role=HA mode=RO

2017-03-10 19:03:36 INFO [7f81c37fe700] Connected with metadata server running on 127.0.0.1:3310

Voila MySQL InnoDB Cluster configuré et prêt à être testé !

L’application doit se connecter au port 6446 (écritures et lectures vers le noeud primaire). Les lectures peuvent également être dirigées vers le port 6447.

Tests

Port de lecture

$ watch -td -n1 "mysql -uroot -P6447 --protocol=tcp -pxxx -BNe'SELECT @@port;' 2> /dev/null"

1	$ watch -td -n1 "mysql -uroot -P6447 --protocol=tcp -pxxx -BNe'SELECT @@port;' 2> /dev/null"

=> affiche : 3320, 3330, 3320, 3330, …

Port d’écriture

$ watch -td -n1 "mysql -uroot -P6446 --protocol=tcp -pxxx -BNe'SELECT @@port;' 2> /dev/null"

1	$ watch -td -n1 "mysql -uroot -P6446 --protocol=tcp -pxxx -BNe'SELECT @@port;' 2> /dev/null"

=> affiche : 3310, 3310, 3310, …

Failover automatique

Grâce à la méthode dba.killSandboxInstance() on peut simuler un crash du serveur primaire et ainsi voir à l’oeuvre le failover automatique du cluster.

Les 2 sessions qui suivent s’exécutent en parallèle:

Session 1

$ watch -td -n1 "mysql -uroot -P6446 --protocol=tcp -pxxx -BNe'SELECT @@port;' 2> /dev/null"

1	$ watch -td -n1 "mysql -uroot -P6446 --protocol=tcp -pxxx -BNe'SELECT @@port;' 2> /dev/null"

Session 2

mysql-js> dba.killSandboxInstance(3310)
The MySQL sandbox instance on this host in 
/.../mysql-sandboxes/3310 will be killed


Killing MySQL instance...

Instance localhost:3310 successfully killed.

mysql-js> dba.killSandboxInstance(3310)

The MySQL sandbox instance on this host in

/.../mysql-sandboxes/3310 will be killed

Killing MySQL instance...

Instance localhost:3310 successfully killed.

=> La session 1 va afficher 3310 puis 3320 après le basculement de la base de données (database failover).

En complément je vous invite à lire :

et regarder la video de lefred : MySQL InnoDB Cluster: MySQL Shell starter guide

Documentation

Thanks for using MySQL!

Filed under:
Group Replication, Haute Disponibilité, InnoDB Cluster by Olivier DASINI

FAQ Webinar MySQL Group Replication

mars 3, 2017

Tags: group replication, Haute disponibilité, InnoDB Cluster, ProxySQL, router

Mise-à-jour: 7 Mars 2017

Le 1er mars dernier, j’ai présenté lors d’un webinar, la technologie de haute disponibilité MySQL Group Replication. On a explosé notre record d’affluence et j’ai été inondé de questions, preuve s’il en faut de votre intérêt, toujours plus important, pour la base de données Open Source la plus populaire au monde.

Je n’ai malheureusement pas été en mesure de répondre à toutes les questions, ce qui me permet aujourd’hui de les centraliser dans cette FAQ. Cette dernière pourra être mise à jour de temps à autres, en fonctions des questions que je récupérerai du terrain.

En complément de cette FAQ

Sur quels OS peut on utiliser MySQL Group Replication ?

MySQL Group Replication est un plugin du serveur MySQL. Il est donc présent sur toutes les plateformes où MySQL 5.7 est disponible :

Linux
Windows
Solaris
OSX
FreeBSD

Liste complete

Est ce que la feature de multi-master nécessitera un update du driver MySQL ou cela sera vu comme un master « standard » d’un point de vu client ?

L’architecture classique MySQL Group Replication est la suivante :

Client + MySQL Router | MySQL Group Replication

Client | ProxySQL | MySQL Group Replication

C’est du moins le cas en mode Single Primary car il faut pouvoir identifier le nœud primaire.

Cette architecture est valable également en mode multi-master (ou multi-Primaire), cependant dans ce cas, vu que tout les membres sont master, il suffit simplement que le driver contiennent la liste des membres.

Est-ce que l’on a une contrainte sur le nombre de nœuds minimum ?

Pas de contraintes par « design ». Cependant, pour qu’un system distribué puisse fournir du failover automatique, il faut minimum 3 nœuds. Cela permet d’éviter le split brain.

A noter qu’il est possible de mettre en place un cluster MySQL Group Replication avec seulement 2 nœuds. Mais en cas d’arrêt non prévu (crash) d’un des membres, l’autre membre n’acceptera pas de nouvelles requêtes. On se trouve alors dans un cas de Network Partitioning, et la partition doit être débloquée selon la procedure inscrite dans la doc.

Par contre si l’un des 2 membres est arrêté « gracieusement » (STOP GROUP_REPLICATION;), l’autre continue de fonctionner normalement.

Dans le cas d’une utilisation multi-master, est-ce que l’on peut ajouter/retirer des nœuds dynamiquement ?

Oui, dans les 2 modes (Single Primary ou multi-master) il est possible d’ajouter et/ou de retirer des nœuds dynamiquement.

Au sein d’un cluster MySQL Group Replication, l’élection d’un nœud master R/W est-il automatique ou manuel ?
Dans le mode Single Primary, lorsque le primaire tombe, comment est choisi le nouveau nœud primaire ?

Le process d’election est automatique et transparent pour le client. Alors pas mal de process se passent « behind the scene » !

Il faut notamment que les autres membres identifient le fait que le nœud ne fasse plus partie du cluster et puis ensuite enclencher le process d’élection du nouveau nœud primaire (le prochain dans la liste ordonnée en fonction de l’UUID de instances du cluster) et aussi désactiver le mode read-only.

Identifier le nœud primaire : https://dev.mysql.com/doc/refman/5.7/en/group-replication-find-primary.html

Exemples:

L’information qui permet de savoir quel nœud est primaire est disponible dans la table performance_schema.global_status :

mysql>
SELECT * 
FROM performance_schema.global_status 
WHERE VARIABLE_NAME='group_replication_primary_member'\G
*************************** 1. row ***************************
VARIABLE_NAME: group_replication_primary_member
VARIABLE_VALUE: 00014115-1111-1111-1111-111111111111

mysql>

SELECT *

FROM performance_schema.global_status

WHERE VARIABLE_NAME='group_replication_primary_member'\G

*************************** 1. row ***************************

VARIABLE_NAME: group_replication_primary_member

VARIABLE_VALUE: 00014115-1111-1111-1111-111111111111

En la joignant avec la table performance_schema.replication_group_members ont obtient un peu plus d’infos:

mysql>
SELECT MEMBER_ID, MEMBER_HOST, MEMBER_STATE
FROM performance_schema.replication_group_members
INNER JOIN performance_schema.global_status ON (MEMBER_ID = VARIABLE_VALUE)
WHERE VARIABLE_NAME='group_replication_primary_member'\G
*************************** 1. row ***************************
MEMBER_ID: 00014115-1111-1111-1111-111111111111
MEMBER_HOST: 192.168.1.11
MEMBER_STATE: ONLINE

mysql>

SELECT MEMBER_ID, MEMBER_HOST, MEMBER_STATE

FROM performance_schema.replication_group_members

INNER JOIN performance_schema.global_status ON (MEMBER_ID = VARIABLE_VALUE)

WHERE VARIABLE_NAME='group_replication_primary_member'\G

*************************** 1. row ***************************

MEMBER_ID: 00014115-1111-1111-1111-111111111111

MEMBER_HOST: 192.168.1.11

MEMBER_STATE: ONLINE

Est-ce que l’on a des limites en terme de taille de base de données pour MySQL Group Replication ?

Le moteur de stockage utilisé avec MySQL Group Replication est InnoDB. Les limites sont donc celles imposées par InnoDB.

Est-ce que l’on peut utiliser MySQL Router avec une replication asynchrone classique ?

MySQL Router 2.0, peut être utilisé avec MySQL Replication : https://dev.mysql.com/doc/mysql-router/2.0/en/

Le plugin Connection Routing fournit 2 modes de « routage »:

Le mode Read-Only : route les requêtes en mode round-robin sur une liste d’instances MySQL.
=> Cela a du sens pour router les requêtes de lectures sur les slaves.

Le mode Read-Write : route les requêtes sur le premier serveur de la liste, si celui si n’est plus accessible, les requêtes sont alors routées sur le suivant de la liste et ainsi de suite jusqu’au dernier.
=> Perso, je ne suis pas trop fan de ce mode pour MySQL Replication.

Sur une configuration multi-master, est il toujours conseillé de mettre en place un offset sur les ID pour chaque node pour éviter les conflits ?

Avec la réplication classique (asynchrone) de MySQL, pour une architecture multi-master en actif/actif (que je ne conseille toujours pas), il est nécessaire de mettre en place différents offsets pour prévenir les conflits liés aux clés primaire en auto increment (details).

Dans le cas de MySQL Group Replication, ce problème se pose également en mode multi-master. C’est pour cela qu’il est géré automatiquement par le cluster. Il est cependant possible de modifier les paramètres manuellement (details).

La modification d’une table est-elle automatiquement répliquée ?

Toutes les modifications certifiées d’un nœud vont être répliquées sur les autres. Cela concerne évidemment les DDL.

A noter que certaines précautions doivent être prisent lorsque vous exécutez des DDL en mode multi-master (details).

Avec 3 nœuds, si on perd 1 nœud et que les 2 autres ne se voient plus, pas de quorum donc plus d’écriture ?

Avec une architecture MySQL Group Replication de 3 nœuds, l’arrêt non prévu (crash) d’1 membre génère la recomposition automatique du cluster avec 2 nœuds ie aucune intervention n’est nécessaire.

Par contre vous vous trouvez dans une situation « inconfortable » et il faut donc re-provisionner un 3ème nœuds ASAP. Parce qu’avec ce cluster de 2 membres, en cas de crash d’un des 2 ou en cas de split-brain, la recomposition automatique du cluster ne peut plus se faire.

On se trouve alors dans le cas du Network Partitioning (les membres n’acceptent plus d’écriture), et la partition doit être débloquée « manuellement ».

A quel format le log binaire doit être positionné ?

Seul le format ROW est valide (binlog-format=ROW)

Peut on contrôler les délais de réplication ?

Le Flow-control permet de maintenir les membres du groupe raisonnablement proches les uns des autres.

A noter que d’autres approches pour contrôler les délais de réplication seront disponibles dans le futur.

Je vous conseille la lecture de cet article qui traite du sujet des performances de MySQL group Replication.

Pourquoi 9 nœuds maximum ?

Selon les résultats de nos tests, jusqu’à 9 membres les performances restent très satisfaisantes. De plus ce nombre assure une très bonne disponibilité car le cluster peut alors gérer automatiquement jusqu’à 4 pertes de nœuds simultanément !

A noter qu’il est toujours possible de rajouter des slaves au cluster MySQL Group Replication.

Est-il possible d’avoir une configuration multi OS par exemple un membre sur Linux et un autre membre sur du Windows ?

Tout à fait !

Je peux par exemple avoir l’architecture MySQL Group Replication suivante :

Nœud A sur Linux
Nœud B sur Windows
Nœud C sur OSX

Quelques liens utiles

Documentation MySQL Group Replication
- https://dev.mysql.com/doc/refman/5.7/en/group-replication.html
MySQL Group Replication: Distributed Recovery behind the scenes
- http://mysqlhighavailability.com/distributed-recovery-behind-the-scenes/
Transaction life cycle explained
- http://mysqlhighavailability.com/mysql-group-replication-transaction-life-cycle-explained/
Group communication behind the scenes
- http://mysqlhighavailability.com/group-communication-behind-the-scenes/
Tous les articles de l’équipe MySQL Group Replication
- http://mysqlhighavailability.com/tag/mysql-group-replication/

Webinar à venir
- https://www.mysql.fr/news-and-events/web-seminars/
Webinars sur demande
- https://www.mysql.fr/news-and-events/on-demand-webinars/

Thanks for using MySQL!

Filed under:
Group Replication, MySQL, ProxySQL by Olivier DASINI

Configurer ProxySQL pour MySQL Group Replication

janvier 11, 2017

Tags: group replication, ha, InnoDB Cluster, proxy, ProxySQL

Cet article s’inspire de HA with MySQL Group Replication and ProxySQL de Frédéric Descamps aka @lefred

Dans un précédent article je vous ai présenté comment déployer un cluster MySQL Group Replication, la nouvelle solution de haute disponibilité de MySQL.

Ce type d’architecture est souvent utilisé avec un composant qui se place entre l’application et le cluster,composant généralement appelé proxy (quelque chose) ou router quelque chose. Dans cet article, je vais vous présenter le meilleur (selon moi) proxy open source du moment : ProxySQL (1.3.2 : la version GA à la date d’écriture).

Le but de cet article est de créer un PoC Solution HA Base de Données Open Source : MySQL Group Replication et ProxySQL.

L’article étant suffisamment long, je ne vais couvrir ni l’installation des nœuds MySQL (5.7.17), ni le déploiement du cluster. Mais comme je suis sympa, voici les ressources nécessaires pour accomplir ces différentes tâches:

Avertissement!
Ce qui suit est effectué par un professionnel… sur son ordi portable 🙂 Toutes les instances sont en local, ~~car je suis fainéant~~ par commodité.
Attention, à ne pas reproduire tel quel en production.

MySQL Group Replication

MySQL Group Replication se présente comme un plugin embarqué dans MySQL à partir de MySQL 5.7.17.

Caractéristiques

Version du serveur : 5.7.17
Version du plugin : 1.0
Nœud 1 : 127.0.0.1 : 14418
Nœud 2 : 127.0.0.1 : 14419
Nœud 3 : 127.0.0.1 : 14420

A partir d’ici, on a donc un cluster MySQL Group Replication de 3 nœuds, installé, déployé et qui fonctionne :

node3 [14420]>
SELECT left(version(),6);
+-------------------+
| left(version(),6) |
+-------------------+
| 5.7.17            |
+-------------------+

SELECT * FROM performance_schema.replication_group_members\G
*************************** 1. row ***************************
CHANNEL_NAME: group_replication_applier
MEMBER_ID: 00014418-1111-1111-1111-111111111111
MEMBER_HOST: localhost
MEMBER_PORT: 14418
MEMBER_STATE: ONLINE
*************************** 2. row ***************************
CHANNEL_NAME: group_replication_applier
MEMBER_ID: 00014419-2222-2222-2222-222222222222
MEMBER_HOST: localhost
MEMBER_PORT: 14419
MEMBER_STATE: ONLINE
*************************** 3. row ***************************
CHANNEL_NAME: group_replication_applier
MEMBER_ID: 00014420-3333-3333-3333-333333333333
MEMBER_HOST: localhost
MEMBER_PORT: 14420
MEMBER_STATE: ONLINE

node3 [14420]>

SELECT left(version(),6);

+-------------------+

| left(version(),6) |

+-------------------+

| 5.7.17 |

+-------------------+

SELECT * FROM performance_schema.replication_group_members\G

*************************** 1. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 00014418-1111-1111-1111-111111111111

MEMBER_HOST: localhost

MEMBER_PORT: 14418

MEMBER_STATE: ONLINE

*************************** 2. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 00014419-2222-2222-2222-222222222222

MEMBER_HOST: localhost

MEMBER_PORT: 14419

MEMBER_STATE: ONLINE

*************************** 3. row ***************************

CHANNEL_NAME: group_replication_applier

MEMBER_ID: 00014420-3333-3333-3333-333333333333

MEMBER_HOST: localhost

MEMBER_PORT: 14420

MEMBER_STATE: ONLINE

Le cluster est en mode single primary, c’est à dire qu’un seul nœud est disponible en écriture (à la fois).

node3 [14420]>
SHOW VARIABLES LIKE 'group_replication_single_primary_mode';
+---------------------------------------+-------+
| Variable_name                         | Value |
+---------------------------------------+-------+
| group_replication_single_primary_mode | ON    |
+---------------------------------------+-------+

SELECT MEMBER_ID, MEMBER_HOST, MEMBER_PORT, MEMBER_STATE
FROM performance_schema.replication_group_members
INNER JOIN performance_schema.global_status ON (MEMBER_ID = VARIABLE_VALUE)
WHERE VARIABLE_NAME='group_replication_primary_member'\G
*************************** 1. row ***************************
MEMBER_ID: 00014418-1111-1111-1111-111111111111
MEMBER_HOST: localhost
MEMBER_PORT: 14418
MEMBER_STATE: ONLINE

node3 [14420]>

SHOW VARIABLES LIKE 'group_replication_single_primary_mode';

+---------------------------------------+-------+

| Variable_name | Value |

+---------------------------------------+-------+

| group_replication_single_primary_mode | ON |

+---------------------------------------+-------+

SELECT MEMBER_ID, MEMBER_HOST, MEMBER_PORT, MEMBER_STATE

FROM performance_schema.replication_group_members

INNER JOIN performance_schema.global_status ON (MEMBER_ID = VARIABLE_VALUE)

WHERE VARIABLE_NAME='group_replication_primary_member'\G

*************************** 1. row ***************************

MEMBER_ID: 00014418-1111-1111-1111-111111111111

MEMBER_HOST: localhost

MEMBER_PORT: 14418

MEMBER_STATE: ONLINE

Pour pouvoir automatiser le routage des requêtes, mais aussi permettre le failover avec ProxySQL nous allons enrichir le schéma sys de MySQL 5.7.17 avec le code ci-dessous (disponible ici également) :

USE sys;

SET GLOBAL log_bin_trust_function_creators=ON;

DELIMITER $$

CREATE FUNCTION IFZERO(a INT, b INT)
  RETURNS INT
RETURN IF(a = 0, b, a)$$

CREATE FUNCTION LOCATE2(needle TEXT(10000), haystack TEXT(10000), offset INT)
  RETURNS INT
RETURN IFZERO(LOCATE(needle, haystack, offset), LENGTH(haystack) + 1)$$

CREATE FUNCTION GTID_NORMALIZE(g TEXT(10000))
  RETURNS TEXT(10000)
RETURN GTID_SUBTRACT(g, '')$$

CREATE FUNCTION GTID_COUNT(gtid_set TEXT(10000))
  RETURNS INT
  BEGIN
    DECLARE result BIGINT DEFAULT 0;
    DECLARE colon_pos INT;
    DECLARE next_dash_pos INT;
    DECLARE next_colon_pos INT;
    DECLARE next_comma_pos INT;
    SET gtid_set = GTID_NORMALIZE(gtid_set);
    SET colon_pos = LOCATE2(':', gtid_set, 1);
    WHILE colon_pos != LENGTH(gtid_set) + 1 DO
      SET next_dash_pos = LOCATE2('-', gtid_set, colon_pos + 1);
      SET next_colon_pos = LOCATE2(':', gtid_set, colon_pos + 1);
      SET next_comma_pos = LOCATE2(',', gtid_set, colon_pos + 1);
      IF next_dash_pos < next_colon_pos AND next_dash_pos < next_comma_pos THEN
        SET result = result +
          SUBSTR(gtid_set, next_dash_pos + 1,
            LEAST(next_colon_pos, next_comma_pos) - (next_dash_pos + 1)) -
          SUBSTR(gtid_set, colon_pos + 1, next_dash_pos - (colon_pos + 1)) + 1;
      ELSE
        SET result = result + 1;
      END IF;
      SET colon_pos = next_colon_pos;
    END WHILE;
    RETURN result;
END$$

CREATE FUNCTION gr_applier_queue_length()
  RETURNS INT
  BEGIN
    RETURN (SELECT sys.gtid_count( GTID_SUBTRACT( (SELECT
Received_transaction_set FROM performance_schema.replication_connection_status
WHERE Channel_name = 'group_replication_applier' ), (SELECT
@@global.GTID_EXECUTED) )));
END$$

CREATE FUNCTION gr_member_in_primary_partition()
  RETURNS VARCHAR(3)
  BEGIN
    RETURN (SELECT IF( MEMBER_STATE='ONLINE' AND ((SELECT COUNT(*) FROM
performance_schema.replication_group_members WHERE MEMBER_STATE != 'ONLINE') >=
((SELECT COUNT(*) FROM performance_schema.replication_group_members)/2) = 0),
'YES', 'NO' ) FROM performance_schema.replication_group_members JOIN
performance_schema.replication_group_member_stats USING(member_id));
END$$

CREATE VIEW gr_member_routing_candidate_status AS SELECT
sys.gr_member_in_primary_partition() as viable_candidate,
IF( (SELECT (SELECT GROUP_CONCAT(variable_value) FROM
performance_schema.global_variables WHERE variable_name IN ('read_only',
'super_read_only')) != 'OFF,OFF'), 'YES', 'NO') as read_only,
sys.gr_applier_queue_length() as transactions_behind$$

DELIMITER ;

SET GLOBAL log_bin_trust_function_creators=OFF;