If you enable parallelism in a For Loop, any shift registers on error wires automatically become error registers, which allow for errors to be merged across all iterations of the For Loop, as illustrated in the previous graphic.

Because you cannot use shift registers in parallel For Loops, you must build error clusters into an array and then merge the errors. Complete the following steps to structure error wires so that you can handle all potential errors.

1. If no iterations of the parallel For Loop execute, which can occur when a previous error exists, the loop does not record any errors you pass into the loop. To avoid this situation, branch the error wire before it enters the parallel For Loop and connect the new segment around the loop to the top input terminal of the Merge Errors node. Using this wire connection strategy ensures that you can track errors that occurred earlier during execution outside of the loop.
2. Enable indexing for the error output tunnel exiting the loop. If you do not enable indexing, only the error information from the final iteration of the loop is maintained.
3. Use the Merge Errors node to capture error information from the code that executed before the loop execution and also each of the loop iterations.