Table of Contents
The Tungsten Dashboard can be connected to any IP address or hostname/Fully Qualified Domain Name (FQDN) of any Tungsten cluster node. Upon establishing the initial connection, the system performs a topology discovery process. This process generates a list of dataservices (clusters) and datasources (nodes) within the Tungsten cluster.
Subsequent connection attempts are made to the node names that were returned during the topology discovery process. These node names must be Fully Qualified Domain Names (FQDNs) or be otherwise resolvable by the Dashboard backend process.
Please make sure that you have hostname resolution for all nodes via /etc/hosts
or DNS set up and tested before trying to add the cluster to the Dashboard. The initial host plus all
other cluster nodes must be reachable from the server hosting the Dashboard. Make sure that /etc/tungsten/tungsten.ini
is configured consistently, and if using short hostnames, then the hostname -f command
should always return the FQDN.
In cases where the node names are not directly resolvable, a mechanism is provided to create a mapping between the node names and their corresponding IP addresses. This mechanism is included as part of the Helm chart or docker-compose installation processes.
To utilize this feature, please refer to the values.yaml
file in the case of a Helm chart
installation, or the docker-compose.yml
file for a docker-compose installation.
The Dashboard uses a custom communication protocol via a TCP socket. Please be aware that launching and connecting
a Dashboard server to a Tungsten Cluster will cause the Dashboard to show up as a router in the cctrl
command, the APIv8 endpoint /api/v8/manager/cluster/topology
, as well as the APIv2 endpoint.
The Dashboard will not behave like the other routers in this list, since it can only communicate via the TCP connection between the cluster and the Dashboard. While the Dashboard will acknowledge any commands sent to it via the TCP connection, it will only act on a very select few of them. Direct TCP connection to the Dashboard server is NOT supported.
Operations executed via the Dashboard are effectively the same as sending REST API requests directly to the cluster. For more information about how to best manage the impact of the Dashboard on the clusters it is monitoring, please see Chapter 9, Best Practices