One of our customer site keep facing ETL failure due to connection error. The error message are either one of the following:
1) java.sql.SQLException: Io exception: Software caused connection abort: recv failed
2) java.sql.SQLException: ORA-06550: line 0, column 0LS-00907: cannot load library unit
3) java.sql.SQLException: The Network Adapter could not establish the connection, java.net.ConnectException: Connection timed out: connect
There is no error on the ETL because when we ask the customer to re-run the batch file, the error did not recur and the job finish successfully. The error did not happen everyday, only occasionally but the reason still unknown.
We suspecting that the connection is the culprit but the customer dismissed this reason as they have another system which running on the same time in another server and the system never fail.
Therefore we are trying to find another point of view to figure out the reason of this ETL failure. We get the GC log from the ETL and analyze that there is a few time that the GC are full. My question is, can the full GC caused network time out or connection lost to the ETL?
How large is your data volume?
Could you please indicate what's talend build version you are using? More information will be helpful for us to address your issue.