Description:
A similar behaviour of Bug# 9282 (4.1.X) can be reproduced in 5.0.13-rc; I know that a fix has been pushed to 4.1.15 but it says nothing about 5.0.X.
It seems that API node lost connection during delete.
Welcome to the MySQL monitor. Commands end with ; or \g.
Your MySQL connection id is 4 to server version: 5.0.13-rc-log
mysql> select count(*) from table;
+----------+
| count(*) |
+----------+
| 971983 |
+----------+
1 row in set (0.00 sec)
mysql> delete from table;
ERROR 2013 (HY000): Lost connection to MySQL server during query
mysql> select count(*) from table;
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect...
Connection id: 1
Current database: test
+----------+
| count(*) |
+----------+
| 971983 |
+----------+
1 row in set (1.02 sec)
(the table was loaded from a backup of 4.1.X cluster)
management node log:
2005-09-28 09:14:56 [MgmSrvr] ALERT -- Node 2: Node 4 Disconnected
2005-09-28 09:14:56 [MgmSrvr] INFO -- Node 2: Communication to Node 4 closed
2005-09-28 09:14:56 [MgmSrvr] ALERT -- Node 3: Node 4 Disconnected
2005-09-28 09:14:56 [MgmSrvr] INFO -- Node 3: Communication to Node 4 closed
2005-09-28 09:14:56 [MgmSrvr] INFO -- Mgmt server state: nodeid 4 freed, m_reserved_nodes 000000
0000000022.
2005-09-28 09:14:56 [MgmSrvr] INFO -- Mgmt server state: nodeid 6 reserved for ip X.Y.Z.W,
m_reserved_nodes 0000000000000062.
2005-09-28 09:14:56 [MgmSrvr] INFO -- Node 2: Node 6 Connected
2005-09-28 09:14:56 [MgmSrvr] INFO -- Node 3: Node 6 Connected
2005-09-28 09:14:56 [MgmSrvr] INFO -- Node 2: Node 6: API version 5.0.13
2005-09-28 09:14:56 [MgmSrvr] INFO -- Node 3: Node 6: API version 5.0.13
2005-09-28 09:14:59 [MgmSrvr] INFO -- Node 2: Communication to Node 4 opened
2005-09-28 09:15:00 [MgmSrvr] INFO -- Node 3: Communication to Node 4 opened
No info or trace in storage nodes logs.
The system and configuration is the same as bug #11776 and bug #12118 (two dual
opterons with 8 Gb each, with gentoo linux)
Of course, I can provide more informations, just let me know.
How to repeat:
Issue a "delete from" on a big ndb table
