Clickhouse Filling the Pit 3: Left Join changed to Right Join, resulting in incorrect statistical results

Recently encountered a big pit, because the ClickHouse large table is associated with a small table, if the large table is placed on the right, the performance will drop rapidly, and it will not even be executed.What I reported here is "Over 16G abnormal", so I naturally thought of putting the big table on the left, and then changing the Left Join to Right Join. This logic is correct in MySQL, but I didn't expect that the results calculated by ClickHouse are actually quite different., stepped on a big pit.I will explain the problem in detail below.

Please see the issue on gitbub: Why LEFT JOIN RIGHT JOIN return different result? How to resolve it? · Issue #14160 · ClickHouse/ClickHouse ·GitHub

The description of this encounter is basically the same as my situation. I use a distributed table and count statistical conditions, and it turns out that the amount of data is allDoubled the number, and carefully checked the reason. I found the association of distributed tables. ClickHouse actually went to each local table to check the association.