Workflow has an activity called try-catch.  You are able to put the particular activity that was failing/throwing errors into the try branch and then activities to handle the failure into the catch branch.

That way, instead of the workflow terminating when the activity has an error it’ll move to the catch branch and continue on.

You could try: search - catch: wait x minutes and search again.

You can also nest try-catch activities by putting additional try-catch activities into the catch branch to try again after the first failure.

Since it sounds like you are already looping and doing the query again, your catch branch may only need a wait activity to delay it a bit before it runs again.