The following are the highlights in this release:
To meet the demand for cost and pricing of Pegasus, we introduce “Capacity Unit” in this release. This concept is equivalent to AWS DynamoDB‘s “Request Unit”, and HBase’s “Capacity Unit”.
Related PR: #318, XiaoMi/rdsn#239, XiaoMi/rdsn#253 Related Docs: TBD Related Issues:
bugfix: too much error logs by write-throttling reject (XiaoMi/rdsn#243)
bugfix: add debug logs for failure detection (XiaoMi/rdsn#256)
bugfix: fix simple_logger gc log file problem (XiaoMi/rdsn#258)
bugfix: http_message_parser (XiaoMi/rdsn#259)
bugfix: pegasus_server: fix set usage scenario problem when open db (#334)
The new perf counters introduced are:
[pegasus.server] ... perf_counter_read_capacity_unit_size = 4096 perf_counter_write_capacity_unit_size = 4096 [pegasus.collector] ... cu_stat_app = stat cu_fetch_interval_seconds = 8 [task.RPC_RRDB_RRDB_PUT] ... profiler::size.request.server = true [task.RPC_RRDB_RRDB_MULTI_PUT] ... profiler::size.request.server = true [task.RPC_RRDB_RRDB_GET_ACK] ... profiler::size.response.server = true [task.RPC_RRDB_RRDB_MULTI_GET_ACK] ... profiler::size.response.server = true