Spark insert CH，Date will shift -1 day

[Guanpx]

Environment

• OS version: centOS 7
• JDK version: 1.8
• ClickHouse Server version:
• ClickHouse Native JDBC version: v2.5.4
• (Optional) Spark version: 2.4.0
• (Optional) Other components' version: N/A

Steps to reproduce

insert data is :
|2021-01| 2021-01-13|2021-05-19 08:00:00|2021-05-13| 1| |2021-05-13 19:21:07|2021-05-13 19:21:46

when use spark insert to CH, that 2021-01-13 change to 2021-01-12 . . . look like this
|2021-01| 2021-01-12|2021-05-19 08:00:00|2021-05-12| 1| |2021-05-13 19:21:07|2021-05-13 19:21:46

the local tz is Asia/Shanghai
the CH tz is Asia/Shanghai

Other descriptions

[pan3793]

Sounds like you are using Spark to read data from CSV and write into ClickHouse.
The data type Date itself should be timezone no-related.
Thus, I guess there are something wrong when you reading data into Spark, can you provide the catalyst schema of the data? And print the data use Spark?

[Guanpx]

the 2021-01-13 type is hive Date

spark df.show
|2021-01| 2021-01-13|2021-05-19 08:00:00|2021-05-13| 1| |2021-05-13 19:21:07|2021-05-13 19:21:46

insert to CH
|2021-01| 2021-01-12|2021-05-19 08:00:00|2021-05-12| 1| |2021-05-13 19:21:07|2021-05-13 19:21:46

[pan3793]

I cannot reproduce the reported issue. If you are still struggling with it, please provide reproducible code.