Fixed replaying bugs found with multimaster load

The replayer did not signal replaying waiters. Added
mysql_cond_broadcast() after replaying is over.

Assertion on client error failed after replay attempt failed due
to certification failure. At this point the transaction does not
go through client state, so the client error cannot be overridden.
Assign ER_LOCK_DEADLOCK to thd directly instead.

Use timed cond wait when waiting for replayers to finish and
check if the transaction has been BF aborted during the wait.

1 files changed, 9 insertions, 2 deletions

diff --git a/sql/wsrep_client_service.cc b/sql/wsrep_client_service.cc
index 62b90cc032f..dd6d7d0bcfb 100644
--- a/sql/wsrep_client_service.cc
+++ b/sql/wsrep_client_service.cc
@@ -260,6 +260,7 @@ enum wsrep::provider::status Wsrep_client_service::replay()
   replayer_service.replay_status(ret);
   mysql_mutex_lock(&LOCK_wsrep_replaying);
   --wsrep_replaying;
+  mysql_cond_broadcast(&COND_wsrep_replaying);
   mysql_mutex_unlock(&LOCK_wsrep_replaying);
   return ret;
 }
@@ -269,9 +270,15 @@ void Wsrep_client_service::wait_for_replayers(wsrep::unique_lock<wsrep::mutex>&
   DBUG_ASSERT(m_thd == current_thd);
   lock.unlock();
   mysql_mutex_lock(&LOCK_wsrep_replaying);
-  while (wsrep_replaying > 0)
+  /* We need to check if the THD is BF aborted during condition wait.
+     Because the aborter does not know which condition this thread is waiting,
+     use timed wait and check if the THD is BF aborted in the loop. */
+  while (wsrep_replaying > 0 && !wsrep_is_bf_aborted(m_thd))
   {
-    mysql_cond_wait(&COND_wsrep_replaying, &LOCK_wsrep_replaying);
+    struct timespec wait_time;
+    set_timespec_nsec(wait_time, 10000000L);
+    mysql_cond_timedwait(&COND_wsrep_replaying, &LOCK_wsrep_replaying,
+                         &wait_time);
   }
   mysql_mutex_unlock(&LOCK_wsrep_replaying);
   lock.lock();