Howard Chu wrote:
Michael Ströder wrote:
Howard Chu wrote:
Michael Ströder wrote:
Now gdb backtrace of hanging test008 with HEAD and compiled with only -O2:
OK, please try current HEAD again. It's reverted to the original epoll behavior (level-triggered) so this type of problem can no longer occur. Also I've found that setting the descriptor to edge-triggered after-the-fact works, so it still prevents the endless hangup reports from happening.
There was only a change to servers/slapd/daemon.c. Is that right?
At that time, right.
Still I see hangs in test008 with HEAD.
Still unable to reproduce it on any of my machines. But, see if the updated connection.c makes a difference.
It works now with ten iterations of test008.
I'm running this on a rather old laptop. Maybe your machines are too fast to reproduce it. ;-)
Ciao, Michael.