Chapter 10 Network Transport Protocol (PMERR) Error Messages

CAUSE: IPCCHECK has been called when no PMERR has occurred. This message is also returned when a connection or socket is closed at the local user's request.

ACTION: None.

CAUSE: The operation is pending. A reply message will be received when the operation has been completed.

ACTION: None.

CAUSE: TCP was unable to send any of the requested data on a partial send request.

ACTION: Either wait for the connection to become writable and then retry the send, or do a waited send.

CAUSE: The data received ends at the urgent data boundary.

ACTION: None. For notification purposes only.

CAUSE: The most recent urgent data boundary has not yet been reached.

ACTION: None. For notification purposes only.

CAUSE: The connection has entered either simplex-in or simplex-out mode.

ACTION: None. For notification purposes only.

CAUSE: The graceful release shutdown for the connection has been completed. The connection has been released.

ACTION: None. For notification purposes only.

CAUSE: The connection has been terminated by IPC.

ACTION: None. For notification purposes only.

CAUSE: TCP has had an internal error which has caused it to process an inbound FIN packet before the open handshake with IPC was complete.

ACTION: This is an internal error. See Appendix A “Submitting an SR ” of this manual. A protocol level trace of TCP header information will enhance resolution of this defect.

CAUSE: A message for a loopback connection in message mode was fragmented.

ACTION: This is an internal error. See Appendix A “Submitting an SR ” of this manual. A protocol level trace including buffer and message level tracing will enhance resolution of this defect.

CAUSE: A call to a Network Type Manager procedure returned an error status.

ACTION: Enable class 2 logging for NS Transport if not previously enabled. Locate matching log instance for Network Type Manager Error. Error message will indicate class 2, subsystem 3, entity TCP, location 100xx. Parameter is NWTM error returned. (Sign bit set indicates error, not set indicates warning.) Consult NWTM error documentation if available, submit an SR with this information if necessary.

CAUSE: A call to a port send message procedure returned an error status.

ACTION: Enable class 2 logging for NS Transport if not previously enabled. Locate matching log instance for send message error. Error message will indicate class 2, subsystem 3, entity TCP, location 101xx. Parameter is ports error returned. (Sign bit set indicates error, not set indicates warning.) Consult MPE/iX operating system error documentation if available, submit an SR with this information if necessary.

CAUSE: TCP was unable to allocate a connection entry. Potential reasons include running out of memory space for resident data structures, table management failure, or TCP internal error.

ACTION: Enable class 2 logging for NS Transport if not previously enabled. Locate matching log instance for resource error if possible. Error message will indicate class 2, subsystem 3, entity TCP, location 102xx. Parameter is table management error returned. (Sign bit set indicates error, not set indicates warning.) Consult MPE/iX operating system error documentation if available, submit an SR with this information if necessary.

CAUSE: An attempt was made to open a connection which already exists.

ACTION: The connection attempt used a set of local and remote socket numbers (sap) and local and remote IP addresses which were identical to the set in use by a currently functioning connection. If you are connecting from a user socket, open a new local socket or allow transport to choose a random sap in the connect call. If you are connecting using transport chosen random saps, the original connector must have marked the local socket for immediate reuse. In this case, retrying the connect should succeed.

CAUSE: TCP was unable to allocate a socket entry. Potential reasons include running out of memory space for resident data structures, table management failure, or TCP internal error.

ACTION: Enable class 2 logging for NS Transport if not previously enabled. Locate matching log instance for resource error if possible. Error message will indicate class 2, subsystem 3, entity TCP, location 102xx. Parameter is table management error returned. (Sign bit set indicates error, not set indicates warning.) Consult MPE/iX operating system error documentation if available, submit an SR with this information if necessary.

CAUSE: An attempt was made to open a socket with a TCP sap number which is already in use by a currently open socket.

ACTION: Close the original socket.

CAUSE: A TCP operation was attempted which conflicts with expected (IPC/BSD sockets) operation in this state.

ACTION: This is an internal error between TCP and the upper layer protocol (IPC or BSD Sockets). You should submit an SR on this defect. Please include the nature of the call being made, and some indication of the current connection state. If possible please include a copy of the program and the input which caused the error to aid resolution of the problem.

CAUSE: A TCP operation was attempted which is in direct violation of (IPC/BSD sockets) /TCP interface rules.

ACTION: This is an internal error between TCP and the upper layer protocol (IPC or BSD Sockets). You should submit an SR on this defect. Please include the nature of the call being made, and some indication of the current connection state. If possible please include a copy of the program and the input which caused the error to aid resolution of the problem.

CAUSE: A receive reply call was made on a TCP connection; however, there was no data pending at the time.

ACTION: This is an internal error between TCP and the upper layer protocol (IPC or BSD Sockets). You may wish to submit an SR for this defect. Please include the nature of the call being made, and some indication of the current connection state. If possible please include a copy of the program and the input which caused the error to aid resolution of the problem.

CAUSE: A receive reply call was made on a TCP connection in message mode; however, there were no messages pending at the time.

ACTION: This is an internal error between TCP and the upper layer protocol (IPC or BSD Sockets). You may wish to submit an SR for this defect. Please include the nature of the call being made, and some indication of the current connection state. If possible please include a copy of the program and the input which caused the error to aid resolution of the problem.

CAUSE: TCP was unable to process a send because it has run out of message queue elements.

ACTION: This is an internal error between TCP and the upper layer protocol (IPC or BSD Sockets). You may wish to submit an SR for this defect. Please include the nature of the call being made, and some indication of the current connection state. If possible please include a copy of the program and the input which caused the error to aid resolution of the problem.

CAUSE: An attempt was made to send data after a simplex-in or graceful release lingering shutdown call had been executed.

ACTION: This is an internal error between TCP and the upper layer protocol (IPC or BSD Sockets). The upper layer protocol should circumvent any further sends before they reach TCP. You may wish to submit an SR for this defect. Please include the nature of the call being made, and some indication of the current connection state. If possible please include a copy of the program and the input which caused the error to aid resolution of the problem.

CAUSE: Data was received on a connection after a shutdown call with how = "graceful release now" (available only in BSD sockets) had been executed.

ACTION: The remote connection half should not be sending data after the local side has issued a "graceful release now shutdown." This is an application timing issue. The application needs to coordinate the connection shutdown or the local side should do a "simple-in shutdown" and continue to receive until the remote graceful release notification is received.

CAUSE: A Netcontrol shutdown was issued. A connection or sockets may have been shutdown due to the network shutdown; or an operation may have been denied because a network shutdown is in progress.

ACTION: Restart the network to be able to establish new sockets and/or connections.

CAUSE: An operation was denied because a NETCONTROL START has not been issued yet (i.e., the transport has not been started.)

ACTION: Issue a NETCONTROL START and retry the operation.

CAUSE: A TCP Connection was aborted either due to a local error or due to a remote connection abortion (i.e., a reset packet was received).

ACTION: Use IPC or BSD error code to discover source of connection abortion.

CAUSE: A TCP connection did not receive any acknowledgment from the remote connection half for a data packet within the configured number of retransmissions.

ACTION: There are many possible reasons for a retransmission time out to occur. Potential problems include severe network bottlenecks, remote system CPU starvation, incorrect or improperly tuned configuration, remote node failure, gateway or network failure, and servers which are not processing new connection requests (or which are becoming backlogged). Check for remote system and network failure first. If there is no problem reaching the remote system, then the problem is most likely a system or network load problem. Check CPU usage on remote system and/or network traffic capacity especially at gateways or over slow (non-LAN) links. Also, if the problem is at connect time, check that the remote server is processing connection requests in a timely manner. Also see the NS 3000/iX NMMGR Screens Reference Manual for a discussion on tuning TCP retransmission parameters.

CAUSE: A TCP connection did not receive any acknowledgment from the remote connection half for a data packet within the configured MAXIMUM TIME TO WAIT FOR REMOTE RESPONSE.

ACTION: There are many possible reasons for a retransmission time out to occur. Potential problems include severe network bottlenecks, remote system CPU starvation, incorrect or improperly tuned configuration, remote node failure, gateway or network failure, and servers which are not processing new connection requests (or which are becoming backlogged). Check for remote system and network failure first. If there is no problem reaching the remote system, then the problem is most likely a system or network load problem. Check CPU usage on remote system and/or network traffic capacity especially at gateways or over slow (non-LAN) links. Also, if the problem is at connect time, check that the remote server is processing connection requests in a timely manner. Also see the NS 3000/iX NMMGR Screens Reference Manual for a discussion on tuning TCP retransmission parameters.

CAUSE: A TCP connection has not received any acknowledgment from the remote connection half within the configured number of connection assurance retries; therefore we assume that the remote system half has experienced an abrupt termination.

ACTION: Check for remote system failure or network failure. If both the network and the remote system are intact, you may wish to readjust the time out parameters, or disable the connection assurance function by configuring a maximum of zero retries.

CAUSE: The operation failed because TCP was unable to resolve a path to the remote system.

ACTION: Check for proper network configuration. If the configuration appears correct, enable Class 5 logging for NS Transport if not already enabled and locate matching logging entry. Logging event will indicate Class 5, Subsystem 3, Entity TCP, Location 104xx, path failure error. Parameter is path resolution error returned. Consult path resolution error listings to diagnose problem.

CAUSE: Excessive retransmissions prompted TCP to test the path to the remote node (an algorithm called active rerouting or negative advice) and path was reported as lost.

ACTION: Suspect gateway or network failure. If necessary, enable Class 5 logging for NS Transport if not already enabled and locate matching logging entry. Logging event will indicate Class 5, Subsystem 3, Entity TCP, Location 104xx, path failure error. Parameter is path resolution error returned. Consult path resolution error listings to diagnose problem.

CAUSE: In an attempt to adjust to a new path, TCP was unable to resegment the currently outstanding outbound data to the new segment size. Most likely this was due to a lack of available buffers in the outbound pool associated with the new link.

ACTION: To prevent future occurrences, increase the buffer pool size of the link which TCP failed over to, or diagnose and prevent the original link failure which caused the fail over.

CAUSE: The remote connection half continued to send data beyond the advertised final (FIN) sequence number.

ACTION: This is a remote protocol violation. Fix or remove the offending remote implementation.

CAUSE: The remote connection half violated IP precedence conventions and sent a packet with an incorrect precedence.

ACTION: This is a remote protocol violation. Fix or remove the offending remote implementation.

CAUSE: A valid SYN packet was discovered within the window of an already established connection.

ACTION: This is a remote protocol violation. Fix or remove the offending remote implementation.

CAUSE: The remote connection half has begun to send data before the connection startup handshake has been completed (i.e. before the SYN/ACK was sent).

ACTION: This is a remote protocol violation. Fix or remove the offending remote implementation.

CAUSE: An attempt was made to send data before the network was started.

ACTION: This is an IPC or BSD sockets internal error. Please submit an SR.

CAUSE: A TCP connection attempted to remove packets from its retransmission queue, but no packets were found.

ACTION: This is an internal error in TCP. Please submit an SR. A protocol or link level trace including TCP header information will facilitate resolution of this problem.

CAUSE: An unexpected escape or trap condition was detected by TCP.

ACTION: This is a software error. Please submit an SR. A detailed description of the network and system state along with a copy of the program and input which uncovered the error may enhance timely resolution of this problem.

CAUSE: An unknown control option was passed to TCP from a call to IPCCONTROL.

ACTION: Identify and remove offending control option.