MySQL Bugs: #64805: StatementImpl$CancelTask occasionally throws NullPointerExceptions

Bug #64805	StatementImpl$CancelTask occasionally throws NullPointerExceptions
Submitted:	29 Mar 2012 18:07	Modified:	22 Feb 2013 0:18
Reporter:	Adam Bradley	Email Updates:
Status:	Closed	Impact on me:	None
Category:	Connector / J	Severity:	S3 (Non-critical)
Version:	5.1.18	OS:	Any
Assigned to:	Alexander Soklakov	CPU Architecture:	Any

Description:
StatementImpl.java:108 references "connection.getQueryTimeoutKillsConnection()".  However, it's possible for "connection" to have been set to "null" by the time this executes.  This causes short NullPointerException stack traces to be printed to the terminal.

How to repeat:
Caused by a race between closing the statement and the query timeout firing, so difficult to intentionally reproduce.

Suggested fix:
Refactor the NullPointerException catches inside of each of the if-branches into a single block that wraps the whole if statement.

Adam, could you refine, do you share Statement between threads?

Not at the level of our application code.  We are using c3p0, I don't know if it shares references to Statements across threads.

I can reproduce this as well, using a case that looks something along the lines of:

JdbcTemplate jdbc = new JdbcTemplate(dataSource);
jdbc.setQueryTimeout(2);
jdbc.execute("select sleep(2);");

I haven't been able to nail down the exact set of circumstances how it occurs though.

mysql connector 5.1.20
spring 3.0.7

Hi Adam,

It looks reasonable to fix as you proposed.

Scott, thanks for repeatable test case!

Added to changelog for 5.1.24: 

The connection.getQueryTimeoutKillsConnection() method could be
called with a null parameter, causing NullPointerException errors and
stack traces. The error depended on a race condition and so occurred
intermittently.