I am facing some unknown issues in my data pipeline. There are few mappings which dumps data from hadoop to teradata one to one using informatica dei. For some of those mappings I am getting below errors:

1. Unexpected parcel kind received
2. Teradata vlan version is invalid
3. Invalid parcel length

This issue occurs suddenly and there is no change we have had done at our end. In those mappings, when I use the split by argument in sqoop to divide the data set over mappers then I am able to load the data. So the root cause is to have a large set of data is the reason of failure.

However, I am very curious to find out how this can happen without any change in the existing platform. What factors can cause this behaviour ??