Hello,
I've read that when DMA failover is configured, the passive DMA is continuously updated. Does this happen at a database or application level?
I want to understand if my Cassandra database for my primary DMA becomes corrupted, will the passive DMA also become corrupted?
Regards
Hi Benjamin,
I would distinguish two types of information synchronized between failover agents.
- DataMiner files - protocols, elements, scripts, etc. This is synchronized at application level (DataMiner handles this process).
- The data written to the database. The way it's synchronized depends on the type of the database.
- MySQL: synchronization is handled by DataMiner, similar to file synchronization.
- Cassandra: When you set up a failover pair with Cassandra, the two database nodes are joined into the cluster. Since that moment, DataMiner just writes data into the Cassandra cluster, and Cassandra ensures the data is synchronized between the two nodes.
Therefore, the answer to your question depends on the way a corruption is introduced into the database. If it's invalid data pushed by DataMiner, then the data will be identical on both database nodes. If it's an issue with a database node, you should be able to retrieve the data from a healthy node.
Files are being synced from main to backup DMA on DataMiner level.
For the database a Cassandra cluster is set up and Cassandra itself is responsible for keeping the data in sync. Using 'nodetool status' the details of that Cassandra cluster can be looked at, but also in the Failover status in Cube you'll see if the Cassandra nodes are up or not.