During smt_create I often get deadlocks resulting in a test timeout locally. This was fine when running in GH actions until recently where the instance type needed scaling up.
The issue appears to be in the Wait call below, and seems directly related to the number of delete workers in the pool. The 1000 default causes issues on my machine but 5 is OK as is 10,000 so something seems very fishy there.
During smt_create I often get deadlocks resulting in a test timeout locally. This was fine when running in GH actions until recently where the instance type needed scaling up.
The issue appears to be in the Wait call below, and seems directly related to the number of delete workers in the pool. The 1000 default causes issues on my machine but 5 is OK as is 10,000 so something seems very fishy there.