Thread: Connection attempt hangs when server is unresponsive
We are using 1.2.1 of DBCP with our tomcat server to pool connections to databases on the wide area. We had an issue today were our tomcat threads started increasing dramatically when we had a Postgres server go away and we found a lot of threads stuck trying to create a datasource: at org.apache.commons.dbcp.BasicDataSource.createDataSource(BasicDataSource.jav a:753) at org.apache.commons.dbcp.BasicDataSource.getConnection(BasicDataSource.java:5 40) at net.iglass.db.JdbcDbAccess.createConnection(JdbcDbAccess.java:122) at net.iglass.db.JdbcDbAccess.createConnection(JdbcDbAccess.java:134) Since createDataSource is synchronized we traced them to the following thread that had them blocked: at java.net.PlainSocketImpl.socketConnect(Native Method) at java.net.PlainSocketImpl.doConnect(PlainSocketImpl.java:333) at java.net.PlainSocketImpl.connectToAddress(PlainSocketImpl.java:195) at java.net.PlainSocketImpl.connect(PlainSocketImpl.java:182) at java.net.SocksSocketImpl.connect(SocksSocketImpl.java:366) at java.net.Socket.connect(Socket.java:516) at java.net.Socket.connect(Socket.java:466) at java.net.Socket.(Socket.java:366) at java.net.Socket.(Socket.java:179) at org.postgresql.core.PGStream.(PGStream.java:58) at org.postgresql.core.v3.ConnectionFactoryImpl.openConnectionImpl(ConnectionFa ctoryImpl.java:77) at org.postgresql.core.ConnectionFactory.openConnection(ConnectionFactory.java: 65) at org.postgresql.jdbc2.AbstractJdbc2Connection.(AbstractJdbc2Connection.java:1 17) at org.postgresql.jdbc3.AbstractJdbc3Connection.(AbstractJdbc3Connection.java:3 0) at org.postgresql.jdbc3.Jdbc3Connection.(Jdbc3Connection.java:24) at org.postgresql.Driver.connect(Driver.java:235) at org.apache.commons.dbcp.DriverConnectionFactory.createConnection(DriverConne ctionFactory.java:37) at org.apache.commons.dbcp.PoolableConnectionFactory.makeObject(PoolableConnect ionFactory.java:290) at org.apache.commons.dbcp.BasicDataSource.validateConnectionFactory(BasicDataS ource.java:877) at org.apache.commons.dbcp.BasicDataSource.createDataSource(BasicDataSource.jav a:851) at org.apache.commons.dbcp.BasicDataSource.getConnection(BasicDataSource.java:5 40) at net.iglass.db.JdbcDbAccess.createConnection(JdbcDbAccess.java:122) at net.iglass.db.JdbcDbAccess.createConnection(JdbcDbAccess.java:134) My question is why this thread did not error out when it could not talk to the server? Could the server have been up when the connection started and went away in the middle causing errors? Could there be something we have overlooked in our conf that would time it out? When the db server came back, our thread count dropped from 200 down to 35 which is normal. Thanks in advance, Woody ---------------------------------------- iGLASS Networks 211-A S. Salem St Apex NC 27502 (919) 387-3550 x813 www.iglass.net
Woody Woodring wrote: > Since createDataSource is synchronized we traced them to the following > thread that had them blocked: > > at java.net.PlainSocketImpl.socketConnect(Native Method) That's an OS-level method, if it blocks and you have not set loginTimeout there is not much the driver can do. Maybe you should set loginTimeout? (It is more expensive as it spawns extra threads to deal with timeouts in this sort of case, though) -O
My question is why this thread did not error out when it could not talk to
the server? Could the server have been up when the connection started and
went away in the middle causing errors? Could there be something we have
overlooked in our conf that would time it out? When the db server came
back, our thread count dropped from 200 down to 35 which is normal.
Look into
1- Setting the jdbc driver's login timeout
2- make sure that the connection attempts are not trying to do long name resolution lookups
3- If you are using DBCP make sure tomcat is configured to remove abandoned connections (removeAbandoned=true) and related parmeters are set to reasonable value.
--
Usama Munir Dar http://linkedin.com/in/usamadar
Consultant Architect
Cell:+92 321 5020666
Skype: usamadar