the major issue here is not BIG vs BIGINT, but the way we store reports. These should have been stored in text/blob field rather than per-line which is extremely slow and it actually put huge stress on the SQL database. I’ve analyzed this here:
We haven’t implemented the changes yet tho. In the meantime, we bumped the primary keys for the most exposed tables (logs, fact_values) sometime in 2017:
You can easily backport this change into your Foreman without upgrading it up for now, just copy the file and perform migration. This should not break your migration when you choose to upgrade later on.