Closed uwiger closed 7 years ago
I added a commit that makes sure t_fail_node()
doesn't actually stop the master node. While that surely is an interesting case, that's what t_master_dies()
does.
Looks good to me.
I've added a few more tweaks, since the t_master_dies()
test case may still time out (though very rarely). Since gen_leader has a 5-second cycle (that may have something to do with it), it seems prudent to wait at least 10 seconds in gproc_dist:sync()
. The test case should of course allow for at least that much as well.
Still not fully ok, but better, so merged anyway.
re. issue #145
Addresses the
case_clause
int_sync_cand_dies()
Increasing overall timeout may help, but that's just a guess.