If running Replication Agent with the PING JCL parameter indicates a failure to connect to Replication Server, and you are using IBM TCP/IP for MVS, follow this procedure to determine the source of the failure.
To determine the source
of the TCP/IP failure
Run the TSO PING command. If the results indicate success, continue to the next step.
Test connectivity to your Replication Server using ISQL or SYBPING from a machine that does not contain DB2 or your Replication Server.
If the connection fails, the problem is on the machine containing Replication Server. Follow troubleshooting procedures for Replication Server (and Open ClientConnect, if necessary). After you correct those problems, repeat Step 1.
If the connection to Replication Server succeeds, continue to Step 3.
Run the PINGRS program for TCP/IP. The JCL for this job is in the PINGRS member, in the hlq.JCL data set.
Make sure that the STEPLIB or JOBLIB DD statements point to the SAS/C runtime library.
Include the SYSPRINT DD statement in the JCL.
Run the Replication Agent with the PING JCL parameter. If it indicates failure, check the following output:
All Open ClientConnect messages. Record message numbers and save the output.
SYSPRINT DD contents for any messages from TCP/IP. Check the message guides for the appropriate product for further explanation. Save the output.
The Replication Server error log on the Replication Server machine. Error or information messages in this log for Replication Agent indicate that the TCP/IP connection succeeded and that there may be a problem with the Replication Server configuration.
Check the JES2 or JES3 system logs for PING messages. Messages generated by RACF, ACF2, and TCP/IP are most common; however, investigate any message that is not security-related.
Remember that MVS tasking can cause messages to appear 10 to 15 seconds before or after the “STARTED” and “ENDED” messages are issued for the job. Save the output. If steps 1 through 4 do not indicate the component(s) that failed, save all output and continue the TCP/IP diagnosis with assistance from your MVS TCP/IP systems programmer.
Verify that the value of the port configurations matches the value of port_number in the interfaces file on the server that contains the primary Replication Server. The port values must be identical in both files.
If either of these values are incorrect, change them and return to Step 1 of “To determine the source of the TCP/IP failure”. If they are correct, proceed to the next step.
Continue testing with the TSO command NETSTAT ALL. The output from the NETSTAT ALL command shows whether the socket is in use by other applications. Use the output from this command to determine whether the port specified in the Replication Agent configuration file and in the interfaces file is in use by another task on z/OS.
According to IBM technical support, when an IUCV socket
starts on the source system, it uses the socket or port number from
the replicate system.
The IP address is resolved in the HOST.ADDRINFO and HOST.SITEINFO data sets. To test these data sets and verify the IP address, issue the TESTSITE command. Save the output from the command.
If this command output fails to show the correct address, the problem may be an incorrect entry in these data sets (the input to these data sets is the HOST.LOCALS data set). Correct this entry, and repeat this procedure starting with Step 1 of “To determine the source of the TCP/IP failure”.
If the command shows the correct address, proceed to the next step.
Run the TSO command TRACERTE specifying the server name. The results of the TRACERTE command show the paths that the IP address takes to be resolved. (This command requires that the issuing user ID be listed in the OBEYFILE on TCP/IP. Contact your TCP/IP administrator to make this change.)
If the output from this command indicates failure, ask your TCP/IP systems programmer to check the routes to the replicate server. This may involve LAN, bridge, and router connections.
If the output from this command indicates success, verify that the IP address is correct and save the output. Continue to the next step.
Execute the traceroute command from the UNIX machine. Specify the MVS server name. This shows the paths that the IP address takes to be resolved.
If the output from this command indicates failure, have the UNIX Systems Administrator check the routes to the MVS server. This may involve LAN, bridge, and router connections.
If the output from this command indicates success, verify that the IP address is correct, and save the output. Continue to the next step.
Ask the TCP/IP systems programmer to run the TRACE ALL and MORETRACE SOCKETS commands.
WARNING! TRACE commands can create a high volume of output. Make sure that your output file is large enough to handle this output. TCP/IP can abend if a data set full condition exists.
Record the following information about the system on which the trace ran:
TCP/IP address space name (job name)
MVS IP address
Primary Replication Server machine IP address and name
Replication Server machine port
TCP/IP version and maintenance level
MVS version and maintenance level
Send all information collected in steps 1 through 11 in electronic format to Technical Support.
Chapter 8, “Configuring TCP/IP for Replication Agent” in the Replication Agent for DB2 UDB Installation Guide for OS/390 for information on editing the SYGWHOST macro
Sybase Replication Server documentation
IBM documentation for TCP/IP for OS/390
