More Reliable Transport Service

A transmitting node initiates the MRTS protocol by setting the NCCIMRTS flag (X‘04’) of the NCCIFEAT field during initial signon. To specify this field, see 2-92. To accept a transmission stream, the receiving node responds by setting the same value in the response signon record.

The More Reliable Transport Service uses the same procedures found in the Less Reliable Transport Service, with the following additional restrictions:

Any R C requires a T C response: EOF is not a valid response.
Only one R C can be sent in a stream. A sending node discards all subsequent records sent until the receiving node responds with a T C record received.
An EOF record has only one correct response: E C.
If the receiving node sends an EOF at the same time it receives an R C, the EOF is ignored.
A receiving node terminates all connection with a sending node that ignores an R C.

For how to code particular protocol records, see the tables below. Note that reason codes are indicatee within the SCB byte for BSC links and the Length byte for SNA links.

Table 1. Transmitter Cancel for SNA and BSC Abnormal Terminations
Link	RCB	Stream ID	SRCB	SCB	LEN
BSC	98-F8 99-F9	SYSIN SYSOUT	40	00	--
SNA	98-F8 99-F9	SYSIN SYSOUT	40	--	00

Table 2. Receiver Cancel for SNA and BSC Abnormal Terminations
Link	RCB	Stream ID	SRCB	SCB	LEN
BSC	B0	SYSIN SYSOUT	98-F8 99-F9	00 C2	--
SNA	B0	SYSIN SYSOUT	98-F8 99-F9	--	00 01

Table 3. EOF Confirm for SNA and BSC Abnormal Terminations
Link	RCB	Stream ID	SRCB	SCB	LEN
BSC	99-F9	--		00	00
SNA	C0	--	98-F8	00	00

Because the operating systems handle R C control records differently, users may follow the LRTS protocol. If JES2 or RSCS are the subsystems at the sending node, the transmitter is held if an R C control record is received. POWER and JES3 drain the NJE jobs that are waiting to be sent and requeue the NJE jobs so that the job can be resent.

Only after the protocol described above has been followed may the sender attempt to transmit a new job in the completed stream. To avoid retransmitting the same job again, operators should place it into a hold state. Under most circumstances, if the receiver ends the job once, it is ended during subsequent attempts to retransmit the job.