Closed jsantell closed 1 year ago
Caused when running multiple tests in rust/noosphere/tests/integration.rs
from the failing cases each first call to sphere_channel().mutable().save(None)
, ultimately calling sphere_context.db_mut().flush()
. The initial tokio::join
in SphereDb<S>::flush
never completes in either test (intermittently).
Due to sled's flush_async()
causing deadlocks. This was fixed in January 2021, but there hasn't been a release since. Using the main
branch of sled fixes the intermittents, as I'm unable to repro the failures over a reasonable amount of iterations.
I've seen this happen occasionally when running tests locally.