When editing your own question, sometimes the initial problem is something completely different to what you expected, and the original question becomes obsolete. How far do you think you should edit and refine a question before simply opening a new one?
Recent example:
My opinion (as stated in the comments on the RAID-specific question you cite) is that if the existing comments and/or answers (if any) on your question would no longer make sense in the context of the modified question, then you should open a new question. Of course, if your old question has no continuing value, then you should also close your old one (if possible) to keep things tidy.
Stackexchange sites are Question and Answer sites.
People put up answers. if the answers no longer relate to the question at hand because the author of the question has changed it then this is unhelpful.
IMHO, Put up a new question for a new question so that others who come and read this in future will have something meaningful to read.
Given the example, I believe that the original should have been edited. The first question was basically "help me with my performance problem". The first question got no answers, but got some comments that might be pointing you in another direction. Whether or not those are right is another matter.
That said, it seems that you've come to the conclusion on your own that the problem is not Samba, so your initial question was asked on a false assumption. Because of that, the original question, as asked, holds no practical value. Instead of abandoning it, I feel that it should be edited to more accurately reflect the situation at hand, since it's about the same underlying performance issue.
