ctdb-locking: Do not reset real-time priority for lock helpers

When using TDB robust mutexes, the kernel wakes waiting processes one
by one, in the priority list order.  To ensure that ctdb lock helper
processes do not starve, lock helper processes need to run at a higher
priority than smbd.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

diff --git a/ctdb/server/ctdb_lock_helper.c b/ctdb/server/ctdb_lock_helper.c
index d831a2d39b32246f94e7f4d23a900e2858d14070..d0a2a891ff59e3b12d75772c9baa593dcb038266 100644 (file)
--- a/ctdb/server/ctdb_lock_helper.c
+++ b/ctdb/server/ctdb_lock_helper.c
@@ -107,8 +107,6 @@ int main(int argc, char *argv[])
                exit(1);
        }
 
-       reset_scheduler();
-
        ppid = atoi(argv[1]);
        write_fd = atoi(argv[2]);
        lock_type = argv[3];