zjuwangg commented on issue #10380: [FLINK-14662]Distinguish unknown CatalogTableStatistics and zero URL: https://github.com/apache/flink/pull/10380#issuecomment-561680478 hi @KurtYoung @bowenli86 @xuefuz I did some verification and the results are as following: 1. Use HiveRunner(which is based on hive 2.3.4) to create a table and insert some data into it, and get the table properties from the HMS. We can just get `totalSize` and `numFiles` 2. Use hive shell to create an empty table and get properties we can get `totalSize`/`numFiles`/`rawDataSize`/`numRows` but all values are 0. Then I load an external file into this table, `totalSize` and `rawDataSize` and not zero but `numRows` and `rawDataSize`are still zero 3. Use hive shell to alter the above table all table stats to '-1' and continue to execute `select` from this table, everything seems well. So my conclusion: 1. HMS may behave differently from express unknown table stats. 2 Seems that set hive table stats to -1 works well for hive.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
