Checkpoint Restart

With this feature, TIBCO MFT Platform Server can take a checkpoint at a user-specified interval for a transfer.

A checkpoint contains information about the transfer such as file pointers, byte count and compressed byte count. In case of a failure, the transfer does not restart at the beginning of the file but starts at the last checkpoint stored. This feature is especially useful in the case where the network connections are slow or the file you want to transfer is large.

To use checkpoint restart, you have to define the following three parameters:
  • TryNumber: specifies a try count for each initiated transfer before it will fail with a permanent error .
  • RetryInterval: determines how long the initiator waits before the next retry of the transfer. By default, this interval is set to 1 minute.
  • CheckPointInterval: defines how often a checkpoint is taken. By default, this interval is set to 1 minute.

For more information, see TryNumber, RetryInterval and CheckPointInterval.

Details of every checkpoint transfer and every initiated transfer that has the TryNumber parameter defined are stored in a Persistent Queue File (PQF) file. This file is stored in the $CFROOT/PQF directory and is deleted once a transfer is completed successfully or has exceeded the TryNumber parameter.

Note the following points when doing a checkpoint transfer:

  • If the initiator cfsend or cfrecv is shut down during a checkpoint restart transfer, CyberResp has to be stopped and restarted so that it can pick up the PQF file to restart that transfer.
  • If a checkpoint transfer fails when UNIX is a responder, the PQF file is not deleted.
  • When CyberResp is running by a non-root user, the transfer can only be restarted by the same non-root user. Otherwise, the following message is displayed:

    Transfer cannot be restarted because CyberResp user is different than Transfer.