Recovering from errors in remote SWAPDUMP
Errors are more likely to occur in a remote SWAPDUMP using TCP/IP than in a local SWAPDUMP, because TCP/IP connections are less reliable than mainframe channels. The operand SWAPIOERR=RETRY is strongly recommended for remote SWAPDUMP, because in many cases it enables FDRPAS to recover from a communications error. With SWAPIOERR=RETRY, FDRPAS almost always recovers if communications are interrupted while a volume is in KEEPACTIVE state (i.e. not actively being copied), and often recovers while a volume is in an update pass (recovery is less likely during the initial copy pass).
NOTIFYERR=userid is also recommended, in order to notify you of errors immediately. The TSO user specified by NOTIFYERR= should be logged on to the LPAR where the SWAPDUMP job is running, so that error messages are displayed as soon as the user presses Enter.
When you become aware that an error has occurred, it is advisable to browse the SYSOUT while the SWAPDUMP is still running, in order to view the surrounding messages and collect more information about the failure. You can do a Find for '**' (two asterisks) to orient to the error messages, since error messages from the FDR system contain '**' following the message number.
If there is an error that causes the SWAPDUMP to fail for any one volume, FDRPAS attempts to process all of the remaining volumes in the current cycle, and then terminates. If you want to interrupt the SWAPDUMP sooner, you can issue a STOP(P) command. The STOP command causes the SWAPDUMP to terminate after completing the volumes that are currently active, instead of continuing to the end of the cycle. When SWAPDUMP with KEEPACTIVE=REPEAT terminates in either of these ways, the I/O intercepts are left in place for all volumes except those that have failed, so that the initial copy pass will not have to be repeated. You should then correct the error, and resubmit the job. Contact INNOVATION for assistance if desired. Any volumes that failed in the earlier run will start over from the initial copy pass; any other volumes will resume with an update pass copying only tracks that have changed since the earlier run.