Does Presto read Hive cause data loss?

[qianmoQ]

Does Presto read Hive cause data loss?

When a Hive task is completed and the end time is 2022-11-16 05:43:36, run Presto to query Hive data.
Parallel query after the end time Hive uses select count(1) to query the total number of data items: 342434654 Data missing 1238
When select count(1) is used again 5 minutes later, the total number of query data is 342435892

After checking, it is found that there is a block in HDFS at the time of 2022-11-16 05:43:36, and the data inside is exactly 1238. It is speculated that this Block is missing when Presto loads HDFS Block!

What configuration do I need to do for this situation? Or is it configured to load HDFS Block?

[willshen]

I don't think Presto/Trino read can ever cause data loss.

For your issue - do you have caching enabled that might have incomplete data at some point of time?

[qianmoQ]

The hdfs file generated after landing is the same as the query time. There is no time interval