After you modify the configurations, you must restart the cluster by using the rolling restart method. This has minor impacts even on a large-size cluster.
- Logs are generated for configuration modifications.
- Subsequent optimization: Some parameter changes can take effect without the need to restart the cluster.
- The following table describes some of the user-configurable parameters. For more information, view the configurations in the console.
|hbase.hregion.majorcompaction||604800000||The default value is 604800000. Unit: milliseconds. You can set the value to 0 to disable automatic major compaction. Subsequent products of ApsaraDB for HBase will support triggering major compaction periodically or within a time period.|
|hbase.ipc.server.callqueue.read.ratio||0||Specifies the percentage of call queues dedicated to read and write requests. By default, there is no differentiation of queues as read or write.|
|hbase.ipc.server.callqueue.scan.ratio||0||Specifies the percentage of call queues allocated for gets and scan requests. By default, all the read queues are used for both get and scan requests.|
|hbase.regionserver.global.memstore.lowerLimit||0.3||The maximum size of all memstores in a region server before flushes are forced.|
|hbase.regionserver.global.memstore.size||0.35||The percentage of JVM heap allocated for memstores. If the percentage reaches a threshold, flush will be triggered.|
|hbase.rpc.timeout||60000||The timeout period for sending Remote Procedure Call (RPC) requests. Unit: milliseconds. You can extend the timeout period.|
|hfile.block.cache.size||0.4||The percentage of JVM heap allocated for block cache. The default value is 0.4, but a higher percentage, for example, 0.5, will help with read latencies if the primary workload is read.|