The data in StackState is divided into three different sets: Elastic Search data, Kafka Topic data, and the StackGraph data. With this much data to store, it is important to have the means to manage it. There is a standard 8 days data retention period set in StackState, that you can configure according to your needs. Besides that, you can also use a StackState CLI command or choose to perform a few manual steps locally on each machine.
The StackState CLI needs access to the Admin API (default port 7071) to issue the command used below.
Using the StackState CLI route takes care of stopping all necessary services, performs the deletion of all topology and telemetry data, and starts StackState up. However, the Kafka topics folder needs to be deleted manually from the StackState server. The Kafka topics folder is located in
/opt/stackstate/var/lib/ and is named
sts graph delete --all
Data retention configuration provides a balance between the amount of stored data, performance, and data availability. By default, the data retention window is set to 8 days. This works in a way that the latest state of topology graph will always be retained; only history older than 8 days will be removed. Find more here.
Please note that the below instructions are valid for a single node installation type. For a two-node installation, you need to stop the service corresponding to the node -
stop stackgraph for a StackGraph node, for example.
To clean StackState manually, StackState and StackGraph services must be stopped first:
systemctl stop stackstate
systemctl stop stackgraph
After above services are stopped you can remove the directory that holds the files:
rm -rf /opt/stackstate/var/lib/*
Start StackState and StackGraph services:
systemctl start stackstate
systemctl start stackgraph