Timeseries / Influx DB Configuration questions?

  • 17 January 2022
  • 0 replies
  • 45 views

Userlevel 4
Badge

Product: SCH All versions

 

Question:

Influx DB has been configured based on the https://www.influxdata.com/blog/influxdb-clustering/

The setup contains 2 Meta Nodes and 3 Data Nodes.

When installing SCH there is a option to enter the time series database details.


These are the questions:

  • Should we enter the Datanodes or Meta nodes host information
  • When you have multiple Meta nodes / Datanodes should we enter the comma-separated URL or load balancer URL to ensure the HA is correctly configured

Answer:

 +   Load balance the data nodes in this case
 +   You can use any load balancer (HAPROXY,F5, Amazon LB, etc)
 +   Present the load balancer URL to   SCH on that screen.

 

• How SCH will communicate to influx dB while having HA configuration? 
You will have both SCHs point to the same influxdb. SCH f5 load balancer and influxdb HA config will take care of figuring out the rest as if you only had one SCH and one Influx.  You dont have to use F5 you can use any load balancer such as HAProxy and others

• In the configuration file for SCH which meta-node details need to be specified? Can we use the Load Balance URL for meta-nodes? 
You do not need to load balance between the meta-notes. Influx like other NoSQL databases is a fully connected mesh network so each node talks to every other node on availability etc. You need to make sure the meta-nodes can talk to each other and the data nodes, but you need the load balancer only on for the data nodes

• How does the meta-node communicate to data nodes? Do we need to specify data-node details on SCH configuration files?
You need to make sure that the nodes can talk to each other per the standard ports required for influx. The timeseries-app.properties only contains the link to the load balancer for the data notes and NOT the meta nodes. 


This topic has been closed for comments