Skip Navigation Links | |
Exit Print View | |
Oracle Solaris Cluster Data Service for MySQL Cluster Guide Oracle Solaris Cluster 3.3 3/13 |
1. Installing and Configuring HA for MySQL Cluster
A. Files for Configuring HA for MySQL Cluster
B. Deployment Example: Installing MySQL Cluster in the Global Zone
C. Deployment Example: Installing MySQL Cluster in a Non-Global Zone
How to Install MySQL Cluster Software
Setting Up the MySQL Cluster Control
How to Configure the Management Server on Both Nodes
How to Configure the Data Nodes on Both Nodes
How to Initialize the MySQL Server
Example Configuration Files for Installation in a Non-Global Zone
config.ini File for Both Nodes to Store in /mgm-data
my.cnf File for the Data Nodes to Store in /ndbd-data
my.cnf File for the First SQL Node phys-schost-1 to Store in /mysql-data
my.cnf File for the Second SQL Node phys-schost-2 to Store in /mysql-data
mysql_config File for the First SQL Node phys-schost-1 to Store in /temp/cluconfig
mysql_config File for the Second SQL Node phys-schost-2 to Store in /temp/cluconfig
mysql_ndb_mgmd_config File for the First Node phys-schost-1
mysql_ndb_mgmd_config File for the Second Node phys-schost-2
mysql_ndbd_config File for the First Node phys-schost-2
mysql_ndbd_config File for the Second Node phys-schost-2
ndbd_shutdown_config File for One Node
phys-schost-1:/ # clresourcegroup create access-rg phys-schost-1:/ # clressharedaddress create -g access-rg \ > -n phys-schost-1:zone1,phys-schost-2:zone2 sa_host_1 phys-schost-1:/ # clresourcegroup online -eM access-rg phys-schost-1:/ # clresourcegroup create -p maximum_primaries=2 -p desired_primaries=2 \ > -n phys-schost-1:zone1,phys-schost-2:zone2 mgm-rg phys-schost-1:/ # clresourcegroup create -p maximum_primaries=2 -p desired_primaries=2 \ > -n phys-schost-1:zone1,phys-schost-2:zone2 ndbd-rg phys-schost-1:/ # clresourcegroup create -p maximum_primaries=2 -p desired_primaries=2 \ > -n phys-schost-1:zone1,phys-schost-2:zone2 mysql-rg phys-schost-1:/ # clresourcegroup set -p rg_affinities=++ndbd-rg mysql-rg
Note - Setting the ++ affinity ensures that on a restart of a single node, the start order of the resources is maintained as set within the resource dependencies.
phys-schost-1:/ # zlogin zone1 mkdir /cluster-pfiles phys-schost-2:/ # zlogin zone2 mkdir /cluster-pfiles
phys-schost-1:/ # clresourcetype register gds
Use the content of mysql_ndb_mgmd_config File for the First Node phys-schost-1 for phys-schost-1 and mysql_ndb_mgmd_config File for the Second Node phys-schost-2 for phys-schost-2.
ID=1 for zone1
ID=2 for zone2
Value for zone1:
CONNECT_STRING=zone1,zone2
Value for zone2:
CONNECT_STRING=zone2,zone1
zone1:/ # ksh /opt/SUNWscmys/ndb_mgmd/util/mysql_ndb_mgmd_register \ > -f /temp/cluconfig/mysql_ndb_mgmd_config -p zone2:/ # ksh /opt/SUNWscmys/ndb_mgmd/util/mysql_ndb_mgmd_register \ > -f /temp/cluconfig/mysql_ndb_mgmd_config -p
Leave the non-global zone on both nodes. Create the resource on one node's global zone, start the mgm-rg resource and verify with MySQL Cluster methods.
phys-schost-1:/ # ksh /opt/SUNWscmys/ndb_mgmd/util/mysql_ndb_mgmd_register \ > -f /temp/cluconfig/mysql_ndb_mgmd_config phys-schost-1:/ # clresourcegroup online -eM mgm-rg phys-schost-1:/ # /usr/local/mysql/bin/ndb_mgm \ > --ndb-connectstring=phys-schost-1-p,phys-schost-2-p -e show phys-schost-1:/ # /usr/local/mysql/bin/ndb_mgm \ > --ndb-connectstring=phys-schost-2-p,phys-schost-1-p -e show
Use the content of mysql_ndbd_config File for the First Node phys-schost-2 for phys-schost-1 and mysql_ndbd_config File for the Second Node phys-schost-2 for phys-schost-2.
ID=3 for zone1
ID=4 for zone2
zone1:/ # ksh /opt/SUNWscmys/ndbd/util/mysql_ndbd_register \ > -f /temp/cluconfig/mysql_ndbd_config -p zone2:/ # ksh /opt/SUNWscmys/ndbd/util/mysql_ndbd_register \ > -f /temp/cluconfig/mysql_ndbd_config -p
Leave the non-global zone on both nodes.
phys-schost-1:/ # ksh /opt/SUNWscmys/ndbd/util/mysql_ndbd_register \ > -f /temp/cluconfig/mysql_ndbd_config phys-schost-1:/ # clresourcegroup online -eM ndbd-rg
Note - Do not try to take the ndbd-rg resource offline until you create and enable the shutdown controller resource.
Use the content of ndbd_shutdown_config File for One Node.
phys-schost-1:/ # ksh /opt/SUNWscmys/ndbd_shutdown/util/ndbd_shutdown_register \ > -f /temp/cluconfig/ndbd_shutdown_config phys-schost-1:/ # clresourcegroup online -e ndbd-rg
Note - From this point, never take offline on all the servers only the ndbd resource. To shut down the ndbd completely, either use the clresourcegroup take offline ndbd-rg command or first disable the shutdown controller resource.
To shut down an ndbd resource on one node only (performing a rolling restart), you can disable it with clresource disable -n phys-schost-1 ndbd-rs. In this case, you should re-enable the resource before you shut down another resource.
For a rolling restart, do not disable the shutdown controller resource. Doing so would lead to a restart of the ndbd without loading data, in which case your database would be unavailable.
Use the content of ha_mysql_config File for One Node.
phys-schost-1:/ # ksh /opt/SUNWscmys/util/ha_mysql_register \ > -f /temp/cluconfig/ha_mysql_config phys-schost-1:/ # clresourcegroup online -eM mysql-rg