I came across the code below in org.apache.activemq.store.jdbc.adapter.DefaultJDBCAdapter.
As part of https://issues.apache.org/jira/browse/AMQ-3288 the doDeleteOldMessages() was changed...
It seems that priorityIterator gets incremented every time doDeleteOldMessages is called which in turn seems to depend on checkpoint interval. This approach looks a bit unusual to me... Yes, even with interval of in 1 sec it will take more than 68 years to
get to the point where it silently overflow and start generating negative priority, but if user will by accident set it 1 millisecond it will happened in 24 days... Why would we need a ticking bomb in the code?... Am I missing something? Is the object instance is short-lived that it is not important?
int priorityIterator = 0;
public void doDeleteOldMessages(TransactionContext c) throws SQLException, IOException {
PreparedStatement s = null;
cleanupExclusiveLock.writeLock().lock();
try {
LOG.debug("Executing SQL: " + this.statements.getDeleteOldMessagesStatementWithPriority());
s = c.getConnection().prepareStatement(this.statements.getDeleteOldMessagesStatementWithPriority());
int priority = priorityIterator++%10;
s.setInt(1, priority);
s.setInt(2, priority);
int i = s.executeUpdate();
LOG.debug("Deleted " + i + " old message(s) at priority: " + priority);
} finally {
cleanupExclusiveLock.writeLock().unlock();
close(s);
}
}
