My recommendation is to look at each "problem" from a risk perspective. How many times has it happened in the past, how many occurrences now, and what's the likelihood it might happen again? And how significant is the potential impact (e.g., on customers, regulatory compliance, cost, ...)? If it's a low frequency/likelihood and low impact, why not just log it for trend analysis. It it happens again then you can reconsider.
If it's a high frequency and/or impact, then do a RCA.
The risk assessment can also be done each time a cause is found, considering whether or not to go deeper.
Of course it will depend on industry context. In pharma, med devices, nuclear industry, etc., RCA will be done on almost everything. But if you're just making frisbees, probably not.