You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@commons.apache.org by "Hu, ZhangHao" <zh...@telenav.cn> on 2010/03/16 10:13:56 UTC
problem in commons-dbcp-1.2.2 + common-pools-1.5.2/1.5.4
Hi All
I am using commons-dbcp-1.2.2 + common-pools-1.5.2 and meet a dead-lock
problem.
The problem is if all connections in the thread pool are occupied, and
other threads are waiting for db connection.
Then if those threads received InterruptedException, then the thread
pool will be dead and can't be used any more.
But commons-dbcp-1.2.2 + common-pools-1.4 can work will with same
setting and same test program.
The new version can't work properly with commons-dbcp-1.2.2? (I also try
commons-dbcp-1.4 + common-pools-1.5.2, it can't work too...)
Any suggestions are appreciated.
The configuration for dbcp is as follow:
dal.connectionProperties=autoReconnect=true;
dal.initialSize=0
dal.maxActive=1
dal.maxIdle=1
dal.minIdle=0
dal.maxWait=-1
dal.validationQuery=select 1
dal.testOnBorrow=true
dal.testOnReturn=false
dal.testWhileIdle=true
dal.timeBetweenEvictionRunsMillis=60000
dal.numTestsPerEvictionRun=3
dal.minEvictableIdleTimeMillis=60000
dal.poolPreparedStatements=true
dal.maxOpenPreparedStatements=20
The test program is as follow:
package db.test;
import java.util.List;
import com.telenav.dal.content.verse.crawler.ISiteDefinition;
import com.telenav.dal.content.verse.crawler.PageCrawlTaskDao;
import com.telenav.dal.content.verse.crawler.SiteDefinitionDao;
public class TestConnection {
public static void main(String[] args) throws Exception {
Runnable t1 = new Runnable() {
@Override
public void run() {
int cnt = 0;
try {
cnt =
PageCrawlTaskDao.getInstance().getCountAllPageTask(1); // a
time-consuming db operation, it will occupy the connection for some time
}
catch ( Exception e ) {
System.out.println("------------------------------------------------");
e.printStackTrace();
}
System.out.println(cnt);
}
};
Runnable t2 = new Runnable() {
@Override
public void run() {
int cnt = 0;
try {
cnt =
PageCrawlTaskDao.getInstance().getCountAllPageTask(1);
}
catch ( Exception e ) {
System.out.println("catch exception: " +
Thread.currentThread().getName());
//e.printStackTrace();
}
System.out.println("thread " +
Thread.currentThread().getName() + " : " + cnt);
}
};
Thread tt1 = new Thread(t1);
tt1.start();
Thread.sleep(2000);
tt1.interrupt();
int i = 0;
while ( i < 20 ) {
Thread tt2 = new Thread(t2, Integer.toString(i));
tt2.start();
Thread.sleep(1000);
if(i == 3)
tt2.interrupt();
i ++;
}
//tt1.interrupt();
Thread.sleep(60000);
System.out.println("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx");
List<ISiteDefinition> sites = null;
while ( true ) {
try {
// The main thread can't get any
connection here.....
sites =
SiteDefinitionDao.getInstance().getAllSiteDefinitions(); // a
non-time-consuming db operation to show the db connection pool is
correct to use
}
catch (Exception e ) {
e.printStackTrace();
}
if ( sites != null ) {
for ( ISiteDefinition site : sites ) {
System.out.println(site.getUrl());
}
}
Thread.sleep(5000);
}
}
}
Best wishes,
Hu Zhanghao
Re: problem in commons-dbcp-1.2.2 + common-pools-1.5.2/1.5.4
Posted by Phil Steitz <ph...@gmail.com>.
Hu, ZhangHao wrote:
> Hi All
>
>
>
> I am using commons-dbcp-1.2.2 + common-pools-1.5.2 and meet a dead-lock
> problem.
>
> The problem is if all connections in the thread pool are occupied, and
> other threads are waiting for db connection.
>
> Then if those threads received InterruptedException, then the thread
> pool will be dead and can't be used any more.
>
>
>
> But commons-dbcp-1.2.2 + common-pools-1.4 can work will with same
> setting and same test program.
>
> The new version can't work properly with commons-dbcp-1.2.2? (I also try
> commons-dbcp-1.4 + common-pools-1.5.2, it can't work too...)
>
> Any suggestions are appreciated.
>
Thank you for reporting this. Unfortunately, this looks like a pool
bug. It is being tracked as POOL-162 in JIRA
(https://issues.apache.org/jira/browse/POOL-162). The only
workaround that I can suggest immediately is to avoid interrupting
threads blocked waiting on the pool, possibly using a positive
maxWait value as an alternative means of controlling blocked waits.
Comments on the ticket are welcome. The fix looks simple. Assuming
I am right on this, we should get this patched ASAP.
Phil
>
>
> The configuration for dbcp is as follow:
>
> dal.connectionProperties=autoReconnect=true;
>
> dal.initialSize=0
>
> dal.maxActive=1
>
> dal.maxIdle=1
>
> dal.minIdle=0
>
> dal.maxWait=-1
>
> dal.validationQuery=select 1
>
> dal.testOnBorrow=true
>
> dal.testOnReturn=false
>
> dal.testWhileIdle=true
>
> dal.timeBetweenEvictionRunsMillis=60000
>
> dal.numTestsPerEvictionRun=3
>
> dal.minEvictableIdleTimeMillis=60000
>
> dal.poolPreparedStatements=true
>
> dal.maxOpenPreparedStatements=20
>
>
>
> The test program is as follow:
>
>
>
> package db.test;
>
>
>
> import java.util.List;
>
> import com.telenav.dal.content.verse.crawler.ISiteDefinition;
>
> import com.telenav.dal.content.verse.crawler.PageCrawlTaskDao;
>
> import com.telenav.dal.content.verse.crawler.SiteDefinitionDao;
>
>
>
> public class TestConnection {
>
> public static void main(String[] args) throws Exception {
>
> Runnable t1 = new Runnable() {
>
> @Override
>
> public void run() {
>
> int cnt = 0;
>
> try {
>
> cnt =
> PageCrawlTaskDao.getInstance().getCountAllPageTask(1); // a
> time-consuming db operation, it will occupy the connection for some time
>
> }
>
> catch ( Exception e ) {
>
>
> System.out.println("------------------------------------------------");
>
> e.printStackTrace();
>
> }
>
>
>
> System.out.println(cnt);
>
> }
>
> };
>
>
>
> Runnable t2 = new Runnable() {
>
> @Override
>
> public void run() {
>
> int cnt = 0;
>
> try {
>
> cnt =
> PageCrawlTaskDao.getInstance().getCountAllPageTask(1);
>
> }
>
> catch ( Exception e ) {
>
> System.out.println("catch exception: " +
> Thread.currentThread().getName());
>
> //e.printStackTrace();
>
> }
>
>
>
> System.out.println("thread " +
> Thread.currentThread().getName() + " : " + cnt);
>
> }
>
> };
>
>
>
> Thread tt1 = new Thread(t1);
>
> tt1.start();
>
> Thread.sleep(2000);
>
> tt1.interrupt();
>
>
>
> int i = 0;
>
> while ( i < 20 ) {
>
> Thread tt2 = new Thread(t2, Integer.toString(i));
>
> tt2.start();
>
> Thread.sleep(1000);
>
> if(i == 3)
>
> tt2.interrupt();
>
> i ++;
>
> }
>
>
>
> //tt1.interrupt();
>
>
>
> Thread.sleep(60000);
>
>
> System.out.println("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx");
>
>
>
> List<ISiteDefinition> sites = null;
>
> while ( true ) {
>
> try {
>
> // The main thread can't get any
> connection here.....
>
> sites =
> SiteDefinitionDao.getInstance().getAllSiteDefinitions(); // a
> non-time-consuming db operation to show the db connection pool is
> correct to use
>
> }
>
> catch (Exception e ) {
>
> e.printStackTrace();
>
> }
>
>
>
> if ( sites != null ) {
>
> for ( ISiteDefinition site : sites ) {
>
> System.out.println(site.getUrl());
>
> }
>
> }
>
> Thread.sleep(5000);
>
> }
>
> }
>
> }
>
>
>
>
>
> Best wishes,
>
> Hu Zhanghao
>
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org