Expert Repair Service configuration reference

Expert repair configuration options

There are some configuration options exposed only for adjustment with the assistance of an expert in DataStax Support.

[repair_service] max_parallel_repairs

The maximum number of subrange repairs to run in parallel. If unspecified or set to 0, the Repair Service calculates the correct number of maximum repairs to run in parallel. Default: 0.

[repair_service] min_throughput

The minimum throughput in bytes needed to calculate parallel repairs. See also num_recent_throughputs. Default: 512. This parameter applies to subrange and incremental repairs only, and is not applicable to distributed subrange repairs.

[repair_service] num_recent_throughputs

The maximum most recent completed repair throughputs used to calculate the average repair throughput, which is then used to determine how many parallel repairs are needed. See also min_throughput. Default: 500. This parameter applies to subrange and incremental repairs only, and is not applicable to distributed subrange repairs.

[repair_service] offline_splits

The minimum number of subrange splits for a node to have per keyspace when falling back to offline splits if the node is unable to communicate with its agent to get more optimal splits. Default: 256. This parameter applies to subrange repairs only.

[repair_service] parallel_tasks_update_interval

Interval at which to recalculate parallel tasks to use for the Repair service in seconds. Default: 120 (2 minutes). This parameter applies to subrange and incremental repairs only, and is not applicable to distributed subrange repairs.

[repair_service] prioritization_page_size

The maximum number of remaining subrange repair tasks for the repair service to evaluate when choosing the next low-impact repair to run. Default: 512. This parameter applies to subrange repairs only.

[repair_service] tokenranges_http_timeout

The timeout in seconds for the HTTP call to the agent to retrieve node token range splits. The default value (30) is deliberately set higher than the default value for DataStax Agents http_timeout in clustername.conf (10). Default: 30. This parameter applies to subrange repairs only, and is not applicable to incremental or distributed subrange repairs.

[repair_service] tokenranges_partitions

Target number of partitions for each range in a subrange repair. This value is used by the Repair Service to create repair tasks, and is set to the max DSE merkle tree depth by default. Default: 1048576.

Note: Do not set tokenranges_partitions higher than default. See Tuning Repair Service for multi-datacenter environments.