Failed server connections
If CICS terminates abruptly, without going through the normal resource manager termination processing for the quasi-reentrant TCB, for example because of a FORCE command or system completion code 40D, an end-of-memory resource manager routine cleans up the client side of the connection.
Because this routine does not run in the CICS region itself, it cannot use the cross-memory connection to notify the server. Therefore, if CICS terminates without carrying out the normal resource manager termination processing for the quasi-reentrant TCB, the server side of the connection might remain active. For coupling facility data tables this might cause problems because the server does not allow the original CICS region to resynchronize after restart if its APPLID is already in use.
From time to time the server checks the client status for each connection, and cleans up the server side of the connection if the client has gone away. This check is done once a minute, and it is also triggered each time a new connection is to be established. Therefore, any connections which have failed are normally cleaned up before the new connection attempts to resynchronize. Because the clean-up processing that terminates the server side of a connection is asynchronous, and might take one or two seconds, the clean-up processing might not be finished in time for the resynchronization from the original CICS region to succeed immediately, but if the resynchronization does not succeed on the first attempt, it should succeed when it is next retried.