We believe we might be observing the behaviour described here:

 

https://www.monetdb.org/blog/monetdb-sql-transaction-management-scheme

 

Specifically “At certain intervals, if the database is idle, the store manager will merge the  updates from the WAL into the primary storage”

 

We perform some heavy updates to the database (we have experimented with putting these in a single transaction or having separate transactions for each, but the behaviour is the same).

The updates are basically dropping tables, recreating them and then a COPY INTO.

 

Once the final load has finished, and has been committed, for around 30 seconds we can query the database, but then for a period of several minutes it basically locks up.

I wonder if there is any way to control this behaviour?

Regards

Steve