summaryrefslogtreecommitdiff
path: root/sql/sql_load.cc
diff options
context:
space:
mode:
authorunknown <guilhem@gbichot3.local>2006-09-05 16:52:05 +0200
committerunknown <guilhem@gbichot3.local>2006-09-05 16:52:05 +0200
commita0b7b2426f01271627d39ae586cd409ce50da69e (patch)
treee3d37dfc52626891f544df507b4fd8fa18d444f0 /sql/sql_load.cc
parenteb0cc3bf8778ca9734c53099a8c30499dcb5a103 (diff)
downloadmariadb-git-a0b7b2426f01271627d39ae586cd409ce50da69e.tar.gz
Fix for BUG#11151 "LOAD DATA INFILE commits transaction in 5.0".
In 5.0 we made LOAD DATA INFILE autocommit in all engines, while only NDB wanted that. Users and trainers complained that it affected InnoDB and was a change compared to 4.1 where only NDB autocommitted. To revert to the behaviour of 4.1, we move the autocommit logic out of mysql_load() into ha_ndbcluster::external_lock(). The result is that LOAD DATA INFILE commits all uncommitted changes of NDB if this is an NDB table, its own changes if this is an NDB table, but does not affect other engines. Note: even though there is no "commit the full transaction at end" anymore, LOAD DATA INFILE stays disabled in routines (re-entrency problems per a comment of Pem). Note: ha_ndbcluster::has_transactions() does not give reliable results because it says "yes" even if transactions are disabled in this engine... sql/ha_ndbcluster.cc: NDB wants to do autocommit if this is LOAD DATA INFILE. For this to not affect all other engines, we move the logic inside ha_ndbcluster. sql/sql_load.cc: This ha_enable_transaction() in mysql_load() forced an autocommit in all engines, while only NDB wants to do that. So we move the logic inside ha_ndbcluster.cc. mysql-test/include/loaddata_autocom.inc: test for engines to see if they autocommit or not in LOAD DATA INFILE mysql-test/r/loaddata_autocom_innodb.result: result for InnoDB (no autocommit) mysql-test/r/loaddata_autocom_ndb.result: result for NDB (autocommit) mysql-test/r/rpl_ndb_innodb_trans.result: result for InnoDB+NDB transactions. Observe that when ROLLBACK cannot rollback the LOAD DATA INFILE in NDB it issues warning 1196 as appropriate. mysql-test/t/loaddata_autocom_innodb.test: test that InnoDB does not autocommit in LOAD DATA INFILE. mysql-test/t/loaddata_autocom_ndb.test: test that NDB does autocommit in LOAD DATA INFIL mysql-test/t/rpl_ndb_innodb_trans-slave.opt: need to tell the slave to use innodb mysql-test/t/rpl_ndb_innodb_trans.test: test of transactions mixing NDB and InnoDB. To see if ROLLBACK rolls back in both engines, with the exception of LOAD DATA INFILE which does not roll back NDB: we see that a LOAD DATA INFILE in NDB commits all what has been done in NDB so far, commits its changes, but does not commit in other engines.
Diffstat (limited to 'sql/sql_load.cc')
-rw-r--r--sql/sql_load.cc5
1 files changed, 0 insertions, 5 deletions
diff --git a/sql/sql_load.cc b/sql/sql_load.cc
index b1ba2e96651..d5faf6ee7e9 100644
--- a/sql/sql_load.cc
+++ b/sql/sql_load.cc
@@ -147,10 +147,6 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
MYF(0));
DBUG_RETURN(TRUE);
}
- /*
- This needs to be done before external_lock
- */
- ha_enable_transaction(thd, FALSE);
if (open_and_lock_tables(thd, table_list))
DBUG_RETURN(TRUE);
if (setup_tables_and_check_access(thd, &thd->lex->select_lex.context,
@@ -394,7 +390,6 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
table->file->extra(HA_EXTRA_WRITE_CANNOT_REPLACE);
table->next_number_field=0;
}
- ha_enable_transaction(thd, TRUE);
if (file >= 0)
my_close(file,MYF(0));
free_blobs(table); /* if pack_blob was used */