Enabling Big SQL High Availability
In Big SQL 4.2, High Availability (HA) has been simplified and can be enabled by installing the same Big SQL Head Node component on a second host in the Ambari cluster. Enabling HA requires the first head node to be set up, so the second will be added after the service has been installed. Unlike in Big SQL 4.1, there isn’t a different component for the second head node or a separate “Enable HA” action.
The first step of enabling HA is to restart the Ambari Server. The installation of the first head node changes the Ambari Server’s configuration, and those changes are processed when the server is restarted. After restarting the server, navigate to the hosts tab in Ambari’s Web UI and select a host that has neither a worker nor a head node installed.
To confirm that the host does not have any Big SQL component installed, you can open the “Add” menu on the host and check that both “Big SQL Worker” and “Big SQL Head” are options.
To enable High Availability, choose the Head node from the host’s “Add” menu. You should see a pop-up window with a new operation for the installation.
When the installation has completed, you will see a head node component on the selected host. The final step to enable High Availability is starting the component. When both head nodes are started, HA is running.
Managing Big SQL High Availability in Ambari
After enabling High Availability, as long as both head nodes are started, one of the two will be the active, or primary head node that is currently a part of your cluster. The other head node will be the standby head node, which becomes the active head node if the other unexpectedly stops.
If you stop the active head node on its own, the standby will take over. To ensure that no head node is active, stop both head nodes or stop the entire service in Ambari.
To check which head node is currently active, open the “Config” tab for Big SQL in the Ambari dashboard. Under bigsql-env, the first parameter (bigsql_active_primary) will list the active head node.
Please note that this parameter is for display purposes only and should not be modified, as doing so may cause the service check and adding a worker node to fail. If you think you may have modified this parameter, manually stop and start each head node to correct the variable.
Disabling Big SQL High Availability in Ambari
The first step to disable HA is to choose which of the two head nodes you would like to remove. Once you have chosen one, navigate to its component list by choosing “Hosts” in the top toolbar, and then choosing the selected host. On that page, one of the listed components will be the Big SQL Head Node. Use the drop-down menu to turn on maintenance mode for that component.
Once maintenance mode has been turned on, you should see a black toolbox icon for the component. If you see the icon, you can proceed to remove the component by using the Decommission command in the same menu. If you don’t see this command, you need to start the component or stop the other head node component. Once you decommission the component, you should see the following pop-up in the Ambari UI:
Wait until the operation successfully finishes running. Once it has done so, you can close the background operations window and use the component menu to delete the component from Big SQL. Double-check that you are deleting the Big SQL head component on the same host that was decommissioned.
For more information about High Availability in Big SQL, please refer to the IBM Knowledge Centre: http://www.ibm.com/support/knowledgecenter/SSPT3X_4.2.0/com.ibm.swg.im.infosphere.biginsights.admin.doc/doc/admin_bigsql_ha.html