# Installing Percona XtraDB Cluster

This is a guide about installing Percona XtraDB Cluster on 3 Debian jessie servers.

## Preparations

We need 3 Debian jessie machines, preferably on the same network:

* db01.example.com
* db02.example.com
* db03.example.com

### Prepare /etc/hosts

On db01:

```
127.0.1.1 db01.example.com db01
127.0.0.1 localhost
10.10.10.2 db02.example.com db02
10.10.10.3 db03.example.com db03
```

On db02:

```
127.0.1.1 db02.example.com db02
127.0.0.1 localhost
10.10.10.1 db01.example.com db01
10.10.10.3 db03.example.com db03
```

On db03:

```
127.0.1.1 db03.example.com db03
127.0.0.1 localhost
10.10.10.1 db01.example.com db01
10.10.10.2 db02.example.com db02
```

### Setup the FW

On db01:

```
ufw allow from 10.10.10.2
ufw allow from 10.10.10.3
```

On db02:

```
ufw allow from 10.10.10.1
ufw allow from 10.10.10.3
```

On db03:

```
ufw allow from 10.10.10.1
ufw allow from 10.10.10.2
```

## Install Percona XtraDB Cluster


### Setup debian repositories

Run these commands on all nodes.

* Download the percona relese .deb:

```
wget https://repo.percona.com/apt/percona-release_0.1-4.$(lsb_release -sc)_all.deb
```

* Install the package:

```
dpkg -i percona-release_0.1-4.$(lsb_release -sc)_all.deb
```

* Update APT:

```
apt update
```

### Install the package

* Install *percona-xtradb-cluster-57*:

```
apt -y install percona-xtradb-cluster-57
```

Enter and verify the *root* user password when prompted.


**Error: We get this in the logs:**

> 2017-08-13T07:54:38.438122Z 12 [Note] Access denied for user 'root'@'localhost' (using password: NO)

## Setup the cluster

### Configure the nodes

* Stop the *mysql* service on all nodes:

```
service mysql stop
```

* Add these in *etc/mysql/percona-xtradb-cluster.conf.d/mysqld.cnf* on *db01*:

```
# XtraDB Cluster
wsrep_provider=/usr/lib/galera3/libgalera_smm.so

wsrep_cluster_name=pxc-cluster
wsrep_cluster_address=gcomm://10.10.10.1,10.10.10.2,10.10.10.3

wsrep_node_name=db01
wsrep_node_address=10.10.10.1

wsrep_sst_method=xtrabackup-v2
wsrep_sst_auth=sstuser:P4$$W0rd

pxc_strict_mode=ENFORCING

binlog_format=ROW
default_storage_engine=InnoDB
innodb_autoinc_lock_mode=2
```

* Use the same sonfiguration for *db02* and *db03* except these lines:

```
wsrep_node_name=db02
wsrep_node_address=10.10.10.2
```

```
wsrep_node_name=db03
wsrep_node_address=10.10.10.3
```

### Bootstarp the first node

* Start the *mysql* service in bootstrap mode:

```
/etc/init.d/mysql bootstrap-pxc
```

* Verify that the service has been initialised:

```
echo "show status like 'wsrep%'" | mysql | grep -E '(wsrep_local_state_uuid|wsrep_local_state|wsrep_local_state_comment|wsrep_cluster_size|wsrep_cluster_status|wsrep_connected|wsrep_ready)'

wsrep_local_state_uuid  a7150f16-7ffc-11e7-a81b-2200ad46db9d
wsrep_local_state       4
wsrep_local_state_comment       Synced
wsrep_cluster_size      1
wsrep_cluster_status    Primary
wsrep_connected ON
wsrep_ready     ON
```

Looks OK!

* Create a user for State Snapshot Tranfer (SST):

```
mysql -p
CREATE USER 'sstuser'@'localhost' IDENTIFIED BY 'passw0rd';
GRANT RELOAD, LOCK TABLES, PROCESS, REPLICATION CLIENT ON *.* TO 'sstuser'@'localhost';
FLUSH PRIVILEGES;
```

### Add more nodes

Now we need to add more nodes in the cluster.

* Start *mysql* on *db02*:

```
/etc/init.d/mysql start
```

* Verify that *db02* has been added:

```
echo "show status like 'wsrep%'" | mysql | grep -E '(wsrep_local_state_uuid|wsrep_local_state|wsrep_local_state_comment|wsrep_cluster_size|wsrep_cluster_status|wsrep_connected|wsrep_ready)'

wsrep_local_state_uuid  cc726457-7ffb-11e7-9879-7bb11ba2758a
wsrep_local_state       4
wsrep_local_state_comment       Synced
wsrep_cluster_size      1
wsrep_cluster_status    Primary
wsrep_connected ON
wsrep_ready     ON
```

The *wsrep_cluster_size* is still '1'. Something is wrong. Running `tcpdump` for ports 3306 and 4567, on *db01*,  reveals that no traffic reaches the first node.

Running `nmap` on *db02* we get:

```
 nmap -sV -p 3306,4567 db01

Starting Nmap 6.47 ( http://nmap.org ) at 2017-08-13 11:55 PDT
Nmap scan report for db01 (10.10.10.1)
Host is up (0.0010s latency).
rDNS record for 10.10.10.1: db01.example.com
PORT     STATE SERVICE VERSION
3306/tcp open  mysql   MySQL (unauthorized)
4567/tcp open  tram
```

Ports are open so no connectivity issues.

References
----------
* https://www.percona.com/doc/percona-repo-config/apt-repo.html
* https://www.percona.com/doc/percona-xtradb-cluster/LATEST/install/apt.html
* https://www.percona.com/doc/percona-xtradb-cluster/LATEST/configure.html
* https://www.percona.com/doc/percona-xtradb-cluster/LATEST/bootstrap.html
* https://www.percona.com/doc/percona-xtradb-cluster/LATEST/add-node.html