Joey
Joey
我也有这个问题啊，不知道怎么解决
cc @rui-mo @PHILO-HE, thanks.
> If the CH failed tests are the ones just fixed by [131f700](https://github.com/apache/incubator-gluten/commit/131f70043004a299127d80ea080d07ff83231041), we can ignore the failure. Yes, we can ignore the failure. ``` [2024-04-01T09:30:12.651Z] - test mergetree table...
@WangGuangxin Perhaps we could create an issue in the upstream Velox community to remove this verification.
@WangGuangxin Any progress of this PR?
Maybe you can use these configurations: * spark.gluten.sql.columnar.filescan = false * spark.gluten.sql.native.writer.enabled = false
@RaoZhiRou-Z I am not very familiar with the CH backend. You need to change the issue title to start with [CH] so that the people related to CH can assist...
@RaoZhiRou-Z Yes. But you can check the actual plan node name of the scan in vanilla Spark, is it `FileScan` or `BatchScan` or `HiveTableScan`. There are three kinds of configuration...
In our test environment, for boost, it took about 11 min before (from 2024-03-21T06:20:47 to 2024-03-21T06:31:13): ``` 2024-03-21T06:20:47.3262778Z + run_and_time install_boost 2024-03-21T06:20:47.3263537Z + install_boost 2024-03-21T06:20:47.3264743Z + github_checkout boostorg/boost boost-1.84.0 --recursive...
@kgpai Could you help to review?