Connection from talend to vertica

vertica_dba

Hey guys, Our developers are complaining that one of their workflows( i believe it is the forst one they fire) always results with below error. and issue gets fixed when they kill the job and reconnect leads me to believe that the connections are getting stale . the runtimecap value is currently set to unlimited for the user and the developers mentioned that they are not using login timeout. Any ideas what is going on here? thanks for your time in advance,Jai Admin WorkFlow's GetAffiliates job is failed due to java.sql.SQLException: [Vertica][VJDBC](100023) Unexpected message type: BindComplete. at com.vertica.io.ProtocolStream.readExpectedMessage(Unknown Source) at com.vertica.dataengine.VResultSet.fetchChunk(Unknown Source) at com.vertica.dataengine.VResultSet.initialize(Unknown Source) at com.vertica.dataengine.VQueryExecutor.readExecuteResponse(Unknown Source) at com.vertica.dataengine.VQueryExecutor.handleExecuteResponse(Unknown Source) at com.vertica.dataengine.VQueryExecutor.execute(Unknown Source) at com.vertica.jdbc.SPreparedStatement.executeWithParams(Unknown Source) at com.vertica.jdbc.SPreparedStatement.executeQuery(Unknown Source)

Prasanta_Pal

It looks like a connection issue, you may try the following: 1. If your connection pool library can perform a "health check" query at a regular interval, this will not only keep the connection alive, but ensure your application is working with a valid connection. 2. A second option is to ensure that "keepalived" daemon is running on the cluster and reduce the keepalive time to a lower value. The following would start sending keepalive probes after 10 minutes of inactivity (vs 2 hours with default settings). # echo 600 > /proc/sys/net/ipv4/tcp_keepalive_time

vertica_dba

thanks Prasanta. can you guve us the steps to start the "keepalived" daemon and then monitor it as well? Thanks,Jai