z/OS Communications Server: SNA Network Implementation Guide
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


RTP transmission stall operator awareness and recovery support

z/OS Communications Server: SNA Network Implementation Guide
SC27-3672-01

If HPR packets are being dropped in the network, an RTP endpoint will request the partner RTP endpoint to retransmit the network layer packet (NLP). There are many reasons why the packet may not be delivered. One common problem is the NLP size which the RTP endpoint is using may be outdated or inaccurate. This situation can lead to an RTP endpoint continuously requesting the same NLP be retransmitted. The sending RTP endpoint honors the request by repeatedly retransmitting the same NLP, but the partner does not receive this NLP. When this occurs, the RTP pipe is considered to be in a transmission stall.

VTAM® has support to provide operator awareness of an RTP pipe experiencing a transmission stall. When a specific NLP is retransmitted six times without being acknowledged by the RTP partner (as being received), and at least 10 seconds has elapsed since the NLP was first retransmitted, VTAM will notify the operator with message IST2245I as follows:
IST2245I XMIT STALL DETECTED FOR RTP puname TO cpname
At this point, VTAM provides RTP transmission stall recovery support for all RTP pipes whose path traverses an Enterprise Extender link. When the RTP endpoint is not directly attached to EE, but the RTP pipe consists of multiple hops with one being a REFIFO hop, VTAM assumes an EE hop is in the path. A REFIFO link is one that indicates that packets flowing across it are not guaranteed to be delivered in order. Recovery support consists one of the following two actions:
  • First, if the HPR connection traverses an one hop EE or a two hop EE connection network, the RTP endpoint immediately lowers its NLP size to the HPR-architected minimum size of 768 bytes. All outbound data is then resegmented to this value to resolve the transmission stall.
  • Second, VTAM starts a path switch between two RTP endpoints. The path switch allows the EE link to provide the RTP endpoints an updated link size for the EE connection. If the stall persists, VTAM issues message IST2246I every 30 seconds until the stall is alleviated. VTAM also lowers its NLP size to the HPR-architected minimum of 768 bytes if a previous path switch did not alleviate the stalled condition. Message IST2246I follows:
    IST2246I XMIT STALL CONTINUES FOR RTP puname TO cpname
If the recovery actions taken by VTAM result in the packets being delivered to the partner, the transmission stall is considered alleviated at which point VTAM issues message IST2247I as follows:
IST2247I XMIT STALL ALLEVIATED FOR RTP puname TO cpname

At this point normal data flow resumes. After 20 minutes of operating with the reduced MTU size, this RTP pipe's original NLP size is restored in an attempt to increase RTP throughput. If the network will still not allow transport of the larger packets, the stall detection and alleviation process will repeat.

If at any point, a transmission stall extends beyond the time limit specified by the HPRSTALL VTAM start option, VTAM automatically initiates termination of the HPR pipe and issues the following message:
IST2253I HPRSTALL TIME EXCEEDED FOR RTP puname TO cpname

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014