Hey Javier,

From an architectural point of view there should be no issues when restarting the DMA whilst cassandra compaction is running. The cassandra is still able to serve queries and thus the DMA should be able to start up.

However during compaction an increase in latency and response time can occur due to compaction being a resource intensive process. This in turn can cause issues with queries timing out or Out of memory scenarios, so some caution is advised rebooting a DMA during compaction.

Rebooting the entire server hardware during compaction should be avoided as this can lead to issues in the cassandra filesystem, which at best will take extra resources to resolve or at worst can corrupt part of the data.