@diox I wonder if the problem is that we are running migrate 2 times... we run it on initialize and then if data_seed is called we flush and run again... maybe flush doesn't remove the migration table rows and so when we run migrate the second time.. nothing happens.

I think "flush" is not strong enough for us in general seems kind of shaky. but need to investigate further.

But the schema is already correct when we reach the seed step, so if the migrations table was empty, migrate would try to apply migrations from the beginning, and fail...

Ultimately I think any data migration we have should be added somehow to the initial data seed.

@bakulf and I debugged a scneario and it could be the case that when running migration the first time we get the new data and then run "flush" and when we run the migration again it won't be applied because it already has been run.