N3T_TCP_Conn_Drop

A number of connections for a TCPIP stack were dropped because of one or more of the following error conditions:

  • Retransmit_dropped: Connections were lost because the retransmit conditional value was exceeded.
  • Probe_dropped: No response was received while sending window probe requests.
  • Alive_dropped: No response was received while sending keepalive probe requests.
  • Tcp_wait_drops: FINWAIT2 timer expiring before receiving the FIN segment.

The TCP/IP dropped connections value is a stack aggregate metric collected for each TCP/IP stack instance. Connections were probably dropped because of network congestion or lost connectivity to the endpoint node.

To monitor the connections dropped because of error conditions, the following metrics are used:

  • z/OS(R) TCP stack layer and availability
  • Response view table
  • Connections dropped

This situation occurs when the number of connections dropped, because of error conditions, is greater than zero (0) for three consecutive intervals. To resolve this problem, use the following steps:

  1. Analyze the individual TCP/IP connections for the stack to determine if there is a pattern based on subnet, or other logical grouping, where the dropped connections are occurring.
  2. Issue a trace route command to determine the most probable routing path and if connectivity to the endpoint node has been lost.
  3. Determine if this path is using a secondary or backup routing path. If it is, identify and fix the problem with the primary path.
  4. Query the routing interfaces on the routing path to determine the number of packets dropped.
  5. Identify the routers along the routing path with the highest numbers of packets dropped.
  6. Validate the router configuration parameters. Probe_dropped and tcp_wait_drops errors might indicate trouble with endpoint application.

Formula: TCP Connections Dropped > 0