remove trimmed gtid from retrieved gtid set

Summary: When raft logs are trimmed as part of TruncateOpsAfter(), the gtid of the trimmed trxs are also removed from executed/logged gtids. However, when binlog files are rotated, the previous gtid written to the new file depends on what files are rotated. During  binlog rotation the prev gtid are the logged/executed gtids of the instance. During relay log rotation prev gtids are the retrieved gtids of the instance. Hence, in addition to trimming logged gtids/executed gtids, we should also trim retrieved gtids. This prevents creation of holes in the instances executed_gtid set as the replicaset goes through multiple promotions over its lifetime.

Reviewed By: anirbanr-fb

Differential Revision: D21375044

fbshipit-source-id: 96fb274cca7

sql/binlog.cc

@@ -12577,6 +12577,21 @@ int trim_logged_gtid(const std::vector<std::string>& trimmed_gtids)
 
   int error = gtid_state->remove_logged_gtid_on_trim(trimmed_gtids);
 
+#ifdef HAVE_REPLICATION
+  if (active_mi && active_mi->rli)
+  {
+    // Remove rli logged gtids. Note that retrieved gtid is not cleared here
+    // since it is going to be updated when the next gtid is fetched
+    error = active_mi->rli->remove_logged_gtids(trimmed_gtids);
+  }
+  else
+  {
+    // NO_LINT_DEBUG
+    sql_print_information("active_mi or rli is not set. Hence not trimming "
+                          "logged gtids from rli");
+  }
+#endif
+
   global_sid_lock->unlock();
 
   return error;

sql/rpl_gtid_state.cc

@@ -456,13 +456,26 @@ enum_return_status Gtid_state::remove_logged_gtid_on_trim(
   const auto& first_gtid= trimmed_gtids.front();
 
   Gtid gtid;
-  gtid.parse(global_sid_map, first_gtid.c_str());
+  if (gtid.parse(global_sid_map, first_gtid.c_str()) != RETURN_STATUS_OK)
+  {
+    // NO_LINT_DEBUG
+    sql_print_error("Failed to parse gtid %s", first_gtid.c_str());
+    RETURN_REPORTED_ERROR;
+  }
+
   rpl_sidno first_sidno= gtid.sidno;
   sid_locks.lock(first_sidno);
 
   for (const auto& trimmed_gtid : trimmed_gtids)
   {
-    gtid.parse(global_sid_map, trimmed_gtid.c_str());
+    if (gtid.parse(global_sid_map, trimmed_gtid.c_str()) != RETURN_STATUS_OK)
+    {
+      // NO_LINT_DEBUG
+      sql_print_error("Failed to parse gtid %s", trimmed_gtid.c_str());
+      sid_locks.unlock(first_sidno);
+      RETURN_REPORTED_ERROR;
+    }
+
     DBUG_ASSERT(first_sidno == gtid.sidno);
     if (gtid.sidno > 0)
     {

sql/rpl_rli.cc

@@ -2225,6 +2225,47 @@ a file name for --relay-log-index option.", opt_relaylog_index_name);
   DBUG_RETURN(error);
 }
 
+int Relay_log_info::remove_logged_gtids(
+    const std::vector<std::string>& trimmed_gtids)
+{
+  DBUG_ENTER("Relay_log_info::remove_logged_gtid");
+  global_sid_lock->assert_some_lock();
+
+  if (trimmed_gtids.empty())
+    RETURN_OK;
+
+
+  Gtid gtid;
+  for (const auto& trimmed_gtid : trimmed_gtids)
+  {
+    if (gtid.parse(global_sid_map, trimmed_gtid.c_str()) != RETURN_STATUS_OK)
+    {
+      // NO_LINT_DEBUG
+      sql_print_error("Failed to parse gtid %s", trimmed_gtid.c_str());
+      RETURN_REPORTED_ERROR;
+    }
+
+    if (gtid.sidno > 0)
+    {
+      /* Remove Gtid from logged_gtid set. */
+      DBUG_PRINT(
+          "info", ("Removing gtid(sidno:%d, gno:%lld) from rli logged gtids",
+           gtid.sidno, gtid.gno));
+
+      if (gtid_set._remove_gtid(gtid) != RETURN_STATUS_OK)
+      {
+        // NO_LINT_DEBUG
+        sql_print_error("Failed to remove gtid(sidno:%d, gno: %lld) from "
+            "rli logged gtids. ", gtid.sidno, gtid.gno);
+        RETURN_REPORTED_ERROR;
+
+      }
+    }
+  }
+
+  RETURN_OK;
+}
+
 void Relay_log_info::end_info()
 {
   DBUG_ENTER("Relay_log_info::end_info");

sql/rpl_rli.h

@@ -299,6 +299,9 @@ class Relay_log_info : public Rpl_info
       ret= 1;
     return ret;
   }
+
+  int remove_logged_gtids(const std::vector<std::string>& trimmed_gtids);
+
   const Gtid_set *get_gtid_set() const { return &gtid_set; }
 
   Gtid_set *get_gtid_set_nc() { return  &gtid_set; }