3

How browsers handle multiple IPs is not only closed, but locked by the mods. I can't see why. It asks for details that are necessary to understand the effects of deciding to use DNS load balancing - a decision that is very much within the hands of professionals managing networks, and so, I would think, perfectly on-topic here. The question is similar to (but significantly less broad and opinion-based than) Is Round-Robin DNS "good enough" for load balancing static content?, which is open and highly-upvoted, so it seems perverse for it to be closed as "Too Broad".

Should this post be reopened? What's the rationale for the lock?

0

2 Answers 2

Reset to default
5

The closed question is too broad because it is insufficiently precise to be answered within the guidelines in the help center. It is asking nine (count 'em, nine) separate questions. The excessive broadness is also nicely demonstrated by the fact that in the four or so years it was open, it gathered nothing that I would consider a complete answer to the question posed. It's also a very nice example of an X-Y question, which isn't enough by itself, but certainly doesn't help.

If you'd like, I'd be willing to switch it over to being closed as off-topic instead, as the question deals with the behaviour of user-level desktop software, which isn't generally considered on-topic.

It was locked because that way we get the nice little "historical significance" marker down the bottom. I'd take it off if I thought there was the slightest chance that the question might get the reopen votes required to override the close, but... yeah.

Jeff's question stays open because it is not too broad, it deals with a sufficiently well-defined problem, and it attracted an answer which addresses the question asked. I strongly disagree with the assertion that you need to know the full and complete behaviour of every browser when presented with multiple A records before you can show that it's a bad idea.

3

I think this is a borderline case, but the too-broad close was likely not about the many subquestions, but the fact that there might very well be significant differences how browsers handle this which would require a list of browsers and their behaviour (which might even be different between operating systems) and we never liked "give me a list of X" type questions.

9
  • And yet the number of major browsers and different possible behaviours is still finite, and giving an answer that carefully describes many different browser-specific behaviours (if they exist) is possible. And: all that information is relevant to Jeff's question at serverfault.com/q/101053/147556. If this question is too broad, then how can Jeff's question not be too broad when any answer to this question would be a relevant subpart of an answer to Jeff's?
    – Mark Amery
    Commented Apr 7, 2018 at 12:38
  • It seems to me that the asker in this case is being penalised not for the actual breadth of their question - which is not really that broad, and is far narrower than other questions on the topic - but for acknowledging its breadth explicitly in the body of the question (by noting that browsers may have different behaviours). I don't think that's right or useful, and think the question should be reopened; an up-to-date answer would be more useful to me than the woolly and subjective takes on DNS round robin that Server Fault does allow!
    – Mark Amery
    Commented Apr 7, 2018 at 12:42
  • And finally, whatever the merits of closure, this answer gives no indication of why the question was locked; I don't see any reason for the decision about whether the question should be closed to be made unilaterally by a mod and then taken out of the hands of the community. Though that is something that only @womble may be able to explain.
    – Mark Amery
    Commented Apr 7, 2018 at 12:44
  • @MarkAmery I removed my other comment, but without any edit from the author I still think its too broad, and I hesitate on the close reason between too broad versus request for learning. I agree with you its an interesting question, but there is no way to get an accepted answer on that question, broad it’s.
    – yagmoth555 Mod
    Commented Apr 7, 2018 at 16:39
  • @yagmoth555 So, do you also think that Jeff's question is too broad? Or serverfault.com/q/60553/147556? Neither you nor Sven have yet addressed my point that any question about whether it's a good idea to use DNS round robin as a failover technique is necessarily broader than this one, because you need to know the answer to this one in order to have an informed view on whether it makes sense to use DNS round robin for failover... and yet the mods have closed this while leaving the other questions that are broader than it alone.
    – Mark Amery
    Commented Apr 7, 2018 at 18:17
  • @yagmoth555 As for the request for learning materials close reason, it clearly doesn't apply here; no off-site materials are being requested and none would need to be provided in order to fully answer.
    – Mark Amery
    Commented Apr 7, 2018 at 18:18
  • 1
    @MarkAmery: Jeffs question in a nutshell is Can this work, with Not really, and here is why being the answer. The other question explicitly asks for a list of browsers and how they behave. I just don't get at all how you come to the conclusion that the former is broader than the other. The same is true for the third question you linked in your comment. Also, I don't see how the question about how different browsers handle this exactly is relevant to system administration - we don't need this amount of detail to understand the limits of what RRDNS can do.
    – Sven Mod
    Commented Apr 7, 2018 at 18:41
  • @Sven it's broader because if all major browsers sensibly fall through to the next IP if they fail to connect to one of the IPs returned in an A record, then RRDNS is a good failover strategy. The top answer at the second question is incomplete because it just skips that point entirely and assumes that it's necessary to update your DNS record to remove IPs when they become unavailable, which needn't be the case. To determine whether the claim that RRDNS doesn't work as a failover system is actually true, you have to know how browsers behave.
    – Mark Amery
    Commented Apr 7, 2018 at 23:02
  • @MarkAmery It’s broad, don’t think only the browser. Infrastructure: dns; bind? windows server dns? dns appliance? public dns? Now server side, iis? apache? session saved server side? Computer; linux? windows or mac? as they all have a differant tcpip stack, which can impact dns caching. Browser side last there. We could write a book with an answer to that question IMO, as too much dependancy affect the answer.
    – yagmoth555 Mod
    Commented Apr 7, 2018 at 23:55

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .