Can it be possible to disable MVCC function?

Hi all,
Can I disable MVCC function ?

My situation is that, I do hadoop batch ingestion on static hdfs files everyday. But the static files may contain multi-day data as the timestamp is from client timestamp.

For example, if I do batch ingestion on path /some/path/20160703/*, I should set intervals 2016-07-01/2016-07-04.

So the next day on 20160704, I should set intervals 2016-07-02/2016-07-05.

This will result to different version segments on every day, and druid by default will only use the newest version segments for querying.

The querying result is not appropriate as it skips old version segments data.

So for my situation, do anyone have any ideas? Or I should do segment merge task ?

Any reply is thankful !

Can you partition your raw data into daily buckets and for each day, only submit a Hadoop index task for that day’s interval?

Or, if you are guaranteed that there is only a 3 day window, you can include say

/some/path/20160702/*
/some/path/20160703/*
/some/path/20160704/*

And then set the interval to just "2016-07-02/2016-07-03" and it will
filter out all of the other days just producing that one day.

I.e. if you cannot clean your raw data layout, you can instead just
include extra data and filter it out on each indexing.

--Eric