This causes an interesting race condition where you can have a class that takes more than the --server-timeout total, but less than --runner-timeout in between results -- if no other clients are finishing classes or checking out classes in the mean time, the server timeout doesn't get updated and the server can bail with 'No client activity for %ss, shutting down.'
I think the fix is just to put a call to self.activity() in TestRunnerServer.report_result.
This causes an interesting race condition where you can have a class that takes more than the
--server-timeout
total, but less than--runner-timeout
in between results -- if no other clients are finishing classes or checking out classes in the mean time, the server timeout doesn't get updated and the server can bail with'No client activity for %ss, shutting down.'
I think the fix is just to put a call to
self.activity()
inTestRunnerServer.report_result
.