APPC/CPI-C Apps Fail After SNA Server Gets UNBIND Type X'0A'

• Microsoft SNA Server, versions 2.11, 2.11 Service Pack 1 (SP1), 2.11 Service Pack 2 (SP2), 3.0, 3.0 Service Pack 1 (SP1), 3.0 Service Pack 2 (SP2)

SYMPTOMS

An APPC or CPI-C application may occasionally fail to establish a session over a particular Local APPC LU/Remote APPC LU/Mode partnership that had previously been used successfully by the application.

Applications may exhibit different symptoms, depending on how the application is written. If the application has timeout, the application will terminate the session after the timeout has been reached. Applications that do not have a timeout may appear to hang (stop responding) while waiting for the Allocate call to complete.

SNA Application APPC API and LU 6.2 Message traces will show something similar to the following when this occurs while using an APPC application:

   APPC Application                 Windows APPC DLL
   ----------------                 ----------------
   TP_STARTED            ->
                                    Open TP Request -> SNA Server
                                    Open TP RSP OK  <- SNA Server
   [MC_]ALLOCATE         ->
                                 Open LU62 Request -> SNA Server

NOTE: These symptoms may also occur for reasons other than these described in this article, so additional troubleshooting may be required.

CAUSE

If independent LU 6.2 session limits over a particular LU/LU/Mode have not been completed or have been reset by the partner (that is, Mainframe), SNA Server will have to renegotiate CNOS. During the renegotiation process, SNA Server will send a BIND to the host over the SNASVCMG mode for the requested Local/Remote APPC LUs. If the partner rejects the SNASVCMG mode BIND with an UNBIND Type x'0A', the SNA Server could unlock the wrong Mode for CNOS negotiations. This causes subsequent CNOS attempts over the requested Local/Remote APPC LUs to fail with a Command Race Reject error. Because CNOS does not complete for the requested Local/Remote APPC LUs, applications requesting sessions over these Local/Remote APPC LUs fail to get an active session.

WORKAROUND

When this problem occurs, the SNA Server service will need to be stopped and restarted to allow applications to get active sessions over the affected Local/Remote LUs.

STATUS

Microsoft has confirmed this to be a problem in SNA Server versions 2.11, 2.11 Service Pack 1, 2.11 Service Pack 2, 3.0, 3.0 Service Pack 1, and 3.0 Service Pack 2. A supported fix is now available, but has not been fully regression tested and should be applied only to systems experiencing this specific problem. Unless you are severely impacted by this specific problem, Microsoft recommends that you wait for the next Service Pack that contains this fix. Contact Microsoft Technical Support for more information.

MORE INFORMATION

The reported instance of this problem appeared to occur after SNA Server received a large number of BIND -RSP or UNBIND messages from the host over a period of a few hours. The following is an example of the events logged by SNA Server in the Windows NT Application Event Log when receiving the BIND -RSP or UNBIND messages:

   Event ID: 18
   Description: APPC session activation failure: BIND negative response or
   UNBIND request received.

   Sense data = 08050000
   Connection = <SNA Server Connection Name>
   LU alias = <Local APPC LU alias>
   PLU alias = <Partner APPC LU alias>
   Mode name = <APPC Mode name>

Keywords          : kbbug2.11 kbbug2.11.sp1 kbbug2.11.sp2 kbbug3.00 kbbug3.00.sp1 snaappc snacpic kbbug3.00.sp2
Version           : WINDOWS:2.11,2.11SP1,2.11SP2,3.0,3.0SP1,3.0SP2
Platform          : WINDOWS
Issue type        : kbbug
Solution Type     : kbfix