Hi,
I would like to run compute (incremental) stats on a kudu table partitioned by id(hash) and time(range). According to the running result on impala-shell and a previous discussion, Impala doesn't support specify partition of incremental stats for non-HDFS table, does it mean Kudu doesn't support partition based incremental stats, or it is doing that automatically? What's the overall behavior of running compute incremental stats on a kudu table?
Also only COMPUTE STATS supports specify columns (discussed here) So what is the preferred way to compute stats (incremental vs non-incremental) on the partitioned kudu table with many columns?
