This change changes the size of the cyclone pool that are set for each integration test. Prior to this change the value was updated recently to 50 and this led to issues running the DAL integration test suite on some larger Linux systems (namely Threadripper workstations with 64 or 128-cores). The test suite issue would manifest as:
thread 'main' panicked at third-party/rust/tokio-1.40.0.crate/src/runtime/scheduler/multi_thread/worker.rs:466:13:
OS can't spawn worker thread: Resource temporarily unavailable (os error 11)
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
A little misleading in that the number of threads per-process and the number of processes both contribute to this failure.
While the update pool size is considerably smaller, the total runtime of lib/dal:test-integration didn't seem to vary by more than about 5 seconds when runs were attempted with different values.
This change changes the size of the cyclone pool that are set for each integration test. Prior to this change the value was updated recently to
50and this led to issues running the DAL integration test suite on some larger Linux systems (namely Threadripper workstations with 64 or 128-cores). The test suite issue would manifest as:A little misleading in that the number of threads per-process and the number of processes both contribute to this failure.
While the update pool size is considerably smaller, the total runtime of
lib/dal:test-integrationdidn't seem to vary by more than about 5 seconds when runs were attempted with different values.