F_JD21AE is an error when HVR tries to establish a connection to the database but doesn’t succeed with the default (or specified) timeout period. The connection to HVR that wants to connect to DB2i succeeds (which may be an agent running on a different machine), but not the actual connection to the database.
F_JD22D3 is an error when HVR submitted some SQL, is expecting a result, but doesn’t retrieve the result in the default (or specified) timeout period.
The environment variable HVR_TCP_KEEPALIVE changes the frequency of TCP keep-alive messages sent between HVR installations. This is often useful if there are timeouts on the network causing connections to be terminated if there is otherwise no communication on the wire, for example if a large volume of data was staged and HVR submitted a command to bulk load this into the destination database and we are simply waiting for the bulk load to return whether data was loaded successfully. It does not appear that in your environment there is an issue like that.
SQL_PREPARE_TIMEOUT defines how much time we allow for a SQL statement to respond after we submitted it. The unit is in seconds, and the default is 60. For the second error HVR reports it could be that setting this variable to a higher value will help, but one would think your database should have responded within 60 seconds.
HVR_ODBC_CONNECT_STRING_ADD is useful to provide extra instructions to the ODBC connection that may be driver-specific.
Hope this clarifies.