When network problems occur, Export and Import will automatically reconnect without a maximum only under certain conditions.
If each side receives a normal TCP FIN termination, such as from a firewall, each side will exit normally and Export will restart without limit and will place the message "restarted by request" in the event log.
Also, if Import shuts down with normal exit status, such as receiving a "connection reset by peer" and Export did not exit with an error, Export will exit the next time it writes to the network and receives SIGPIPE.
The SIGPIPE is handled internally and is invisible, and again there is no maximum.
However, if either side exits due to a TCP abort or other unrecoverable error, Export and Import will exit with error status.
This means that sp_cop will use the same restart policy that it does with other SharePlex processes, which is to stop restarting after the 4th failure.
This is done in Export/Import to stop potential infinite loops of crash/restart.
Some parts of the decision whether to process restart as an error are platform-dependent.
Some OS platforms and firewall products handle firewall termination as an abort, which would trigger the sp_cop error response and the maximum 4 failures.
To help detect/prevent errors or idle timeout, set the parameter SP_XPT_KEEPALIVE to 1.
Export sends a "keep alive" message every 60 seconds to Import if no data is being sent.
To set the parameter:
sp_ctrl> stop export
sp_ctrl> set param SP_XPT_KEEPALIVE 1
sp_ctrl> start export
Refer to the SharePlex reference Guide for more details on the SP_XPT_KEEPALIVE parameter.
