vendredi 16 février 2018

How to deal with internet drops

This is the situations. It seems to be caused by unreliable internet connections, but it could also be that the 3rd party has bad DB locks and then takes too long to respond.

  1. We call a 3rd party API to e.g. to place an order from a voucher number.
  2. There is a timeout, or internet timeout, error.
  3. The call actually PASSED, but we never received a success response.
  4. The operator does not use our retry functionality, but closes the screen and abandons the process. According to the 3rd party, we placed the order successfully, but we have no local record of this due to the unexpected failure.
  5. Later the operator opens the screen again and repeats the process, but now they get an error stating that the order/voucher was already completed.

We've been thinking about ways to deal with this, but none of them solves the problem completely. For example, the "retry" option helps, but if the operator closes the screen then it is once again beyond our control.

Another thing to note, most errors from the 3rd party are legitimate errors, and we should rightfully not continue with the order. This timeout situation is a special case, and it is becoming a support nightmare.

Aucun commentaire:

Enregistrer un commentaire