Monitoring data source connection pool usage

background

Last one 《 The data source connection pool is not closed Could not open JDBC Connection for transaction》 Analyze the source code MyBatis The framework and Spring Transaction management automatically turns off the data source connection pool , It solves the problem that the connection pool of online data sources is full .
But one may be solved and the next , Now you need a connection pool that can monitor the usage of the data source 、 It's almost full of alarm schemes .

Scheme 1

What we use here is Druid data source , Other data sources can use the following scheme 2 .
Druid The data source itself has a filter, After getting the data source connection （ call getConnection Method ） Will call all in turn filter.
So , We can add one when the data source is initialized filter, Used to view the current connection pool , The code is as follows ：

List<Filter> filters = new ArrayList<Filter>();
filters.add(new FilterAdapter(){
    
	@Override
	public DruidPooledConnection dataSource_getConnection(FilterChain chain, DruidDataSource dataSource, long maxWaitMillis) throws SQLException {
    
		int activeCount = dataSource.getActiveCount();
		if (activeCount > (dataSource.getMaxActive() * 0.8)) {
    
			log.error("Druid Warning,the current active count of Druid Connection Pool is " + activeCount + "!");
		}
		return chain.dataSource_connect(dataSource, maxWaitMillis);
	}
});
dataSource.setProxyFilters(filters);

Here, when the current number of connections is greater than the maximum number of connections 80% Will print an error log , Of course, other alarm codes can be added again .

Option two

Scheme 2 can be applied to general data sources , There is no limit to Druid data source . The core idea is , Write an inherited data source subclass , Rewrite the method to get the connection , The code that implements monitoring before obtaining the connection , The sample code is as follows ：

DruidDataSource dataSource = new DruidDataSource(){
    
	protected Logger log = LoggerFactory.getLogger(BaqiDataSource.class);
	@Override
	public DruidPooledConnection getConnectionDirect(long maxWaitMillis) throws SQLException {
    
		int activeCount = this.getActiveCount();
		if (activeCount > (dataSource.getMaxActive() * 0.8)) {
    
			log.error("Druid Warning,the current active count of Druid Connection Pool is " + activeCount + "!");
		}
		return super.getConnectionDirect(maxWaitMillis);
	}
};
// The following initializes the connection pool 
dataSource.setUrl("xxx");
dataSource.setUsername("xxx");
dataSource.setPassword("xxx"));
dataSource.setDriverClassName("xxx"));
...

Conclusion

Scheme I is applicable to Druid data source , Scheme 2 is applicable to all data sources , It can be used according to the situation .