galera-swarm-lxc-ansible/DOC.md

Relevant Info

For information regarding the tools used for this virtual lab with useful relevant links please read README.md

For information regarding the initial installation of all the necessary components please read INSTALL.md

General command-line information for virtual lab usage

Command for checking the docker node status by default from inside the manager node (which is the same as the host machine): docker node ls

Command for using and attaching to LXC container (e.g. to 'worker1'): lxc-attach --name worker1.

One possible limitation of the deployment process with docker swarm, is that static ip assignments are missing in docker swarm, and so the deployment process is based on the overlay networks, the inclusion of which inside the service deployment adds a degree of environmental complexity for the Galera cluster setup, especially since we use a custom stack deployment for the mariadb-galera-swarm image.

For checking the service status this command is suitable with the output from below:

docker service ls

zwueb8s59c0j        stack_anode         replicated          1/1                 colinmollenhour/mariadb-galera-swarm:latest
cl7qlf0hmntp        stack_bnode         replicated          1/1                 colinmollenhour/mariadb-galera-swarm:latest

For checking the status of the Linux Containers see the following command: lxc-ls --fancy

NAME    STATE   AUTOSTART GROUPS IPV4                               IPV6
worker1 RUNNING 0         -      10.0.3.100, 172.17.0.1, 172.18.0.1 -
worker2 RUNNING 0         -      10.0.3.101, 172.17.0.1, 172.18.0.1 -

Galera cluster debugging

For checking the current size of the Galera cluster - meaning all the current nodes connected to the cluster - there exist some useful commands, which should be executed inside the docker container cluster nodes (docker exec -it {container_id} sh), e.g inside the first MariaDB server node:

mysql -u root -e "SHOW GLOBAL STATUS LIKE 'wsrep_cluster_size%';"

In case the result of this command is 1, it means that only one MariaDB server is running inside the cluster, which is the server from which the cluster is bootstraping from (see also inside the file /var/lib/mysql/grastate.dat if this line exists: safe_to_bootstrap: 1 to confirm that this is the case).

• Manual Healthcheck (inside LXC containers): docker exec {container_id} healthcheck.sh

Also, a new database, named e.g "database" and defined inside the "stack.yaml" deployment file, should initially appear inside the Galera cluster on all the MariaDB nodes:

MariaDB [(none)]> show databases;
+--------------------+
| Database           |
+--------------------+
| database           |