jigold
jigold
https://github.com/hail-is/hail/blob/main/batch/test/test_batch.py#L925-L954
Sorry. Must have messed it up in a different way while trying to fix it. See all of the failed PR batches such as this one: https://batch.hail.is/batches/8125943/jobs/208
This is the issue. I ran both queries in the same transaction: ``` mysql> SELECT job_group_self_and_ancestors.batch_id, job_group_self_and_ancestors.ancestor_id, -> JSON_OBJECTAGG(resources.resource, quantity * GREATEST(COALESCE(rollup_time - start_time, 0), 0)) as resources -> FROM...
I might see the issue. The group by in the test queries doesn't actually do a group by like I wanted because there's no aggregation...
Tests are all finally passing!!!!!!!!
Tests are all passing again.
On Azure, one of the tests timed out with 500 responses from the server. I'll need to debug in GCP, but the PR queue is long right now.
I took out the transaction in the migration as we'll need those triggers to be committed even in the event of the migration failing part of the way through changing...
@danking Can you take another look? The only thing I didn't address is the `Phenotypes -> List[Phenotype], VariantChunks -> List[VariantChunk]`. I don't want to rip it out yet in case...
Just realized that I didn't entirely make this parallel enough yet. Will do tomorrow.