sorry if title is not too descriptive, but I have noticed after upgrading to 3.1.2 (from 3.0.5) that I periodically get 50-60 copies of the following query running at the same time:
DELETE FROM core_item_markers WHERE item_member_id > 0 AND item_is_deleted=1
I understand that this is a periodic process that is supposed to be run from time to time to clean up old data. Looking at the code this is being done in __myDestruct of classItemMarket and is run every 300 seconds.
The problem is that there is a race condition in that on heavy board, while update is being done, multiple other clients detect that update is required and you end up in 50 processes doing exactly the same job. I guess some sort of lock should be introduced, indicating that someone is already updating the cache, so that other processed do not try to.