Some days ago, I asked in Troubleshooting for biber for guidelines what to do/check if biber does not produce any output. The occasion to ask this question was of course that I had such a situation myself. However, I already suspected that my particular problem was an easy issue that I just did not check yet. Just to have a concrete example (and to get my problem fixed) I added the description of my particular situation to the question. As suspected, it was an easy problem: There was an old version of biber somewhere in the OS path. This was discovered very quickly through the comments. Then the question was closed as off-topic since my particular problem was due to old software. But the question itself was never restricted to that particular issue (it was explicitly focussed on general guidelines from the very beginning). I answered the comments left for the reason of closure, I edited the question, but there was no reaction to my comments or edits. So I'm asking here on Meta whether this question is actually considered off-topic and why.
Troubleshooting questions are quite welcome but the issue has always been how to answer them. The usual pattern is that a series of comments, none of which can really be viewed as an answer, are posted and we end up with the user saying 'Ah this solves it'. However, normally this comment thread is tied to the issue for the questionner, and it's unlikely that any other user will have the same issue, or at least that they are unlikely to read the question and any answer that could be constructed and use it without having to ask again. At the same time, there are an ever-expanding set of questions that come down to 'you've got an old version of X somewhere' or 'this was true in an old version but it's already been fixed so you should update'.
The aim of the StackExchange network is (broadly) to build up Q&A info that is valuable to multiple people. Posting answers that only help one person as they are very 'localised' is therefore not something that's been popular. (It can also look a bit like 'rep harvesting' as the solution is already sorted and the gain for other users is not clear.) The convention has thus been that troubleshooting questions normally get closed once a solution has been worked out. Historically, we closed such cases as 'too localized' but the network got rid of that reason, and the 'Powers' favour calling such issues off-topic. As none of the other reasons fit any better, we've tending to use that though I think its fair to say a better reason would be preferable. (Such things are network-wide so are not up to the TeX site to decide on.)
In the case of 'troubleshooting Biber' the problem with a general question is I think that it's unlikely to be possible to cover everything and you have to wonder if a new user would find it (and find it useful). For the common example of the cache issue, I'd imagine they're likely to search for the weird error (
recode_data.xml not found in ...), which is in the title of the 'master' question we have on that. Moreover, general questions tend to be best asked deliberately rather than starting from a single issue. So if you feel that a general Biber debugging question is needed, probably think carefully about how to phrase is then post it.