Configuration of ThinLinc for HA operations

In this section, we describe how ThinLinc is configured for high availability.

Installation of a new HA cluster

In this section, we will describe how to set up a new HA cluster. In the examples we will use a primary node with the hostname tlha-primary and IP address 10.0.0.2, a secondary node with the hostname tlha-secondary and IP address 10.0.0.3, and a resource IP address of 10.0.0.4 with the DNS name tlha.

  1. Begin by installing ThinLinc as described in Installing the ThinLinc server on both nodes.

  2. Both nodes in the HA cluster must have the same SSH host key. Copy /etc/ssh/ssh_host_* from the primary host to the secondary host, and restart sshd on the secondary host.

  3. Install and configure the system-level high-availability software, for example the software provided by the ClusterLabs project, which can be found at https://clusterlabs.org. This and other high-availability software may also be provided as part of your distribution, so check for the solution which best fits your requirements before proceeding.

  4. Configure the system’s high-availability software to watch the status of the other machine via the network, and to enable the resource IP address 10.0.0.4 on the active node. The machine with the hostname tlha-primary should normally be active.

  5. Configure each agent server to allow privileged operations both from tlha-primary and tlha-secondary:

    $ sudo tl-config '/vsmagent/allowed_clients=tlha-primary tlha-secondary'

    Also, set the /vsmagent/master_hostname to the DNS name of the HA interface:

    $ sudo tl-config /vsmagent/master_hostname=tlha

    Restart each agent service (vsmagent.service) after changing the configuration values.

  6. Verify operations of the master service (vsmserver.service) on both nodes. Make sure you can start the service properly on both hosts, and connect to the respective hosts. It should be possible to connect, using tlclient, to both tlha-primary and to tlha-secondary.

    Both nodes should be configured with the same subcluster configuration.

    Warning

    It is very important that 127.0.0.1 is not in the list of agent servers of any subcluster. If the master servers are also acting as agent servers, their unique hostnames or IP addresses must be added to /subclusters/<name>/agents instead of 127.0.0.1. The reason for this is that 127.0.0.1 will be a different server based on which master is currently active.

  7. After verifying that normal ThinLinc connections work as intended when using both the primary and the secondary master’s hostname, it is time to enable HA. This is done by setting /HA/enabled to 1, and by specifying the nodes in the cluster in /HA/nodes. For example:

    $ sudo tl-config /HA/enabled=1
$ sudo tl-config '/HA/nodes=tlha-primary.example.com tlha-secondary.example.com'

    Configuration should be identical on both nodes. Restart the master service (vsmserver.service) on both nodes after configuration.

  8. If the master service can’t safely determine which of the two nodes in /HA/nodes is the remote node, and which is the local node, it will start without HA enabled, and log a message. If this happens, validate your hostname and DNS setup. One of the entries of /HA/nodes must match the local machine. Either the resolved IP of one of the entries in /HA/nodes must match the local IP, or one entry must exactly match the local hostname as returned by uname -n.

  9. Once HA has been configured, tests should be performed in order to confirm that the failover works as expected. This can normally be done by simply removing the network cable from the primary node, and ensuring that the secondary node then takes over. Check also that any active ThinLinc sessions have been synchronized from the primary to the secondary node, and that logging in to such a session on the secondary node succeeds once the primary node has been disabled.

Your ThinLinc HA cluster is now configured! When sessions are created, changed or deleted on the currently active node, this information will be synchronized to the other node. If the other node has to take over service, its copy of the session data will be up-to-date, and it can start serving new requests instantly. When the primary node comes up again, the secondary node will resynchronize with the primary node.

Reconfiguring an existing ThinLinc installation into HA mode

If you have an existing ThinLinc installation and want to eliminate the single point of failure (the single master server), the procedure is very much like the procedure for installing a new HA cluster.