Symptom:
Unhandled monarch error on the root actor, hostname=6c5bb643aff4, PID=28054 at time 2026-06-24 19:20:04.525046:
The actor <root>.<test_rdma.ParameterClient client_gpu{'gpus': 0/1}> and all its descendants have failed:
the process this actor was running on failed: Killed(sig=15)
that actor only exists in the test_proc_mesh_rdma test.
The test itself already reports "passed" before this message is printed, so it is likely that something isn't shut down properly and is still observing actors which should be dead.
Symptom:
that actor only exists in the test_proc_mesh_rdma test.
The test itself already reports "passed" before this message is printed, so it is likely that something isn't shut down properly and is still observing actors which should be dead.