When supervising remote actors I can cause a supervisor failure when
remote nodes fail one after another and
the supervisor shutdowns and restarts children.
This can be demonstrated if in test/test_dist_errors.jl the calls to sleep(4) get reduced or eliminated. The problem occurs mostly on CI where test runs get executed more slowly than on a local machine.
The problem has been detected in development, issue #24 : "Detect and handle node failures"
When supervising remote actors I can cause a supervisor failure when
This can be demonstrated if in
test/test_dist_errors.jlthe calls tosleep(4)get reduced or eliminated. The problem occurs mostly on CI where test runs get executed more slowly than on a local machine.The problem has been detected in development, issue #24 : "Detect and handle node failures"