I had a customer using Debian 64 Bit binaries downloaded from MySQL, experiencing hangs. Symptoms have been:

• INSERT hangs
• Replication hangs (SQL_Thread shown as running, but Exec counter not moving)
• Server cannot be shutdown, except with kill -9

This is Bug #8555, Debian Bug 314408, a problem with NPTL in libc6, version 2.3.2.ds1-22. Upgrading to a newer, UNSTABLE version of libc6 resolved the problem. The system is now running without problems at 1200 qps (4000 qps peak) in 64 Bit mode, using the MySQL AMD 64 bit download for 5.0.18.

The behaviour of SYSDATE() has changed between MySQL 4.1 and 5.0. The manual indicates: Within a stored routine or trigger, SYSDATE() returns the time at which it executes, not the time at which the routine or triggering statement began to execute. This differs from the behavior for NOW(). Obviously, this was implemented internally by having SYSDATE() ignore SET TIMESTAMP, as can be easily tested.


Continue reading "sysdate() weirdness"

MySQL (MyISAM) exhibits strange and little known (but documente) behaviour when dealing with compound keys where a part of the compound is AUTO\_INCREMENT.

This is as everybody knows:

mysql> create table standard ( name varchar(20), i integer not null auto_increment primary key );
Query OK, 0 rows affected (0.18 sec)

mysql> insert into standard ( name ) values ( "kris" ), ("azundris"), ("kris"), ("azundris");
Query OK, 4 rows affected (0.00 sec)
Records: 4 Duplicates: 0 Warnings: 0

mysql> select * from standard;
+----------+---+
| name | i |
+----------+---+
| kris | 1 |
| azundris | 2 |
| kris | 3 |
| azundris | 4 |
+----------+---+
4 rows in set (0.00 sec)


Continue reading …