The Cassandra documentation recommends that a repair job is run every week.
But one thing that is not clear: Do I need to run nodetool repair on/for each node; or does nodetool repair issued on a single node repair the entire cluster?
Running nodetool repair on a single node certainly leads to observable repair activities on all nodes as far as I can see from opscenter.
I agree that the docs are not entirely clear on that. But further down (from the once-a-week recommendation) are these two points:
The hard requirement for routine repair frequency is the value of gc_grace_seconds. Run a repair operation at least once on each node within this time period. Following this important guideline ensures that deletes are properly handled in the cluster.
And...
Use caution when running routine node repair on more than one node at a time and schedule regular repair operations for low-usage hours.
"Running nodetool repair on a single node certainly leads to observable repair activities on all nodes"
Yes, that's because the repair operation needs to look at similar token ranges on other nodes to ensure proper replication/repair.
