1

Edit (by MadHatter, not the original poster): before posting a response to this question, please read the entire thing, including the comments thread at the bottom. The OP deleted the original question (after grokking the SF way, as I understand from the comments), but voretaq7 undeleted it to encourage discussion. The OP has explicitly requested that it be deleted in the comments that follow; it may therefore not represent the OP's current position.

The actual question follows.


I have found the serverfault mods to be unnecessarily administrating the closing of questions.

If you look at my history of questions, I have 700 points on stackoverflow, almost exclusively accumulated by asking questions.

It is my experience that the environment of serverfault is unlike any other site of the stackexchange network.

I have had multiple questions closed because the mods apparently:

  1. Don't understand that the question is specifying details implicitly with regards to what it is asking.
  2. Ie, that the question is specific to a particular instance of failure and a person who knows the answer would easily understand the context of the question.

This is inherent to all questions and certainly to questions that are relevant to semi obscure server configuration.

  1. It is quite vexing to be patronized because the mods of this particular exchange seem to think that because they do not immediately understand the context of the question with regards to a particular configuration (or that a beginner should or would), starting from the very basics, that the question does not provide enough context or does not exhibit all of the troubleshooting required to get to that point.
  2. Please stop enforcing arbitrary and excessive requirements to provide an introduction to the question and specify the entire troubleshooting efforts. Most answers to questions and certainly most ANSWERS TO QUESTIONS WITH REGARDS TO ALREADY IMPLEMENTED CONSTRUCTS OF FUNCTIONALITY THAT ARE CONFIGURABLE, are readily apparent to persons skilled in the art.
  3. This exchange should have an even laxer meta with regards context because the questions often deal with well defined constructs of knowledge, ie they are in fact configurable at runtime.

Anyway the point is:

  1. Stop closing questions and wait for a person to come along that knows the answer or wait until a bounty is available or wait until the questioner replies with an answer.
  2. It is very vexing to spend the time writing a question and then getting an inquisition from mods who are not intimately aware of the configuration implying that you have not provided enough information. It is a waste of time and the purpose of the question is await assistance as a convenience.

Frankly, this is infuriating and quite frankly it is abhorrent that serverfault exists separate from stackoverflow because the topics discussed are intimately related to program execution at runtime and would benefit from the faster response times generated by a larger audience.

  • 11
    None of your questions were actually closed by mods. The community voted on them, that is kind of a big difference. – faker Dec 5 '13 at 9:26
  • @MetaChrome Normally I don't undelete mSO questions when the OP deletes them (I consider them "withdrawn" and thus resolved), but I think yours actually warrants an answer - we really don't want you to go away unhappy! If after I've posted my answer you still want to delete this discussion I'll respect that choice. – voretaq7 Dec 5 '13 at 16:14
  • I'm actually fine. I didn't realize the explicit meta of this site with regards to professional sys admins and I understand the difficulty of enforcing said meta. I think that it is a fine meta, but it sometimes difficult to ascertain if a does in fact provide enough context or not. Sometimes they do sometimes they don't. In general, I think I'll try to keep more general (l)unix questions to unix.stackexchange – MetaChrome Dec 5 '13 at 16:17
  • In the sense that, if a question is coming from a position of examining alternatives from a logical standpoint of a sys admin or if it's just another, plop down a question of the top of your head. I think serverfault's explicit meta is actually VERY valuable. – MetaChrome Dec 5 '13 at 16:24
  • 4
    I think this question ought to be deleted because it doesn't really offer a proposal to maintain a certain meta and the specified elements are context specific. In addition, it's title is an incorrect portrayal of moderators once you understand the explicit goals of the site. Discussions of the explicit goals of the site are best left to the already opened discussions. Eg, maybe the explicit ought to be changed (which I don't really have any input on, because I could see the usefulness of maintaining explicit order) but that isn't relevent to the context of this question. – MetaChrome Dec 5 '13 at 17:10
  • 3
    As an actual moderator, I never saw the question being discussed before it was closed. My first reaction would have been to kick it over to Unix & Linux where open-ended questions about kernel internals and Unix history are more welcome. Server Fault is about making things work and solving concrete problems. Also, very closely related: How can I ask better questions on Server Fault? – Michael Hampton Dec 5 '13 at 19:03
  • 1
    You mean like Bill the Lizard. He deletes questions all day long: sergworks.wordpress.com/2012/09/26/why-stackoverflow-sucks/… . I hope he won't delete this comment :) – WeGoToMars Sep 5 '14 at 13:32
18

quite frankly it is abhorrent that serverfault exists seperate from stackoverflow because the topics discussed are intimately related to program execution at runtime and would benefit from the faster response times generated by a larger audience.

Woah there pardner, back up the horse!

This is really the only thing in your post I vehemently disagree with - and I felt so strongly about it I wanted to put it in its own answer.

What you're basically suggesting in that little snippet is that you want developers to get to play sysadmin and have mob rule -- merge the two sites and let the devs shout down the ops folks (because by glory they do out-number us by about 2 orders of magnitude), but that ignores a core reason to have dedicated system administrators: System Administrators exist to keep developers honest.
We're the reason that application isn't allowed to run as root (or Local System, or an Enterprise Admin…).

Stack Overflow and Server Fault represent two sides to the computing coin - the developers (who write code and will do anything to make it run) and the operators (who take that code, integrate it into a stable environment, and will do anything to make sure that environment stays stable). Ideally both sides work together closely, but the roles, goals, and mindsets are by their nature very different, and which group you want to hear from depends on what your goals are. That's why Server Fault was split off from Stack Overflow to begin with.

  • To clarify once again, I deleted the question because I read the discussions with regards to the explicit meta of this site. I see the value of having said explicit meta, and the value the major contributers get from it or the efforts of said contributers. I dont really have the motivation to discuss the abstract paradigms of system administration or see the point of doing so. It's entirely possible that the paradigm you specify is entirely correct. – MetaChrome Dec 5 '13 at 17:44
  • 1
    @MetaChrome: It is rather magnanimous of you to allow for the fact that a situation that exists in nearly every single IT organization larger than 10 employees. – Scott Pack Dec 5 '13 at 18:05
  • 3
    @MetaChrome I know you seem to agree on reconsideration - you're bright enough to read through Meta. I'm more concerned about the next person who comes along and doesn't take the time to understand why things are split the way they are - those are the people who think "devops" means just hiring a bunch of coders and giving them root on machines to do as they please (with predictable results) – voretaq7 Dec 5 '13 at 18:07
  • 1
    Merge SO and SF? Ulp. – tombull89 Dec 5 '13 at 18:07
  • Thanks Scott. I'll be sure to let you know of my further treatises on the manner. – MetaChrome Dec 5 '13 at 18:09
  • The problem comes down to the fact that, stackexchange has claimed to relegate all questions with regards to configuration of server software to serverfault while the community or the explicit meta of serverfault wants to maintain these question in the context of the "method" of system administration. Now that's an actual question worthy of discussion. If you really think that every question regarding configuration or executables can only be dealt with by system administrators or in the context of the the method of system adminstration, that's debateable, hence the friction. So superuser? – MetaChrome Dec 5 '13 at 18:13
  • I mean the explicit meta of serverfault is in conflict with itself by nature of the fact that there is a misunderstanding of the purpose of the site because of its popularity. You can't possibly believe that you aren't going to get non sys admins posting questions on here unless there is a lot of education about the fact that there is an expectation that the audience is sys admins. – MetaChrome Dec 5 '13 at 18:18
  • 1-media-cdn.foolz.us/ffuuka/board/a/image/1371/30/… Just put this in the question input form with an explanation regarding the meta and I guaruntee you'll never get another dumb question. – MetaChrome Dec 5 '13 at 18:22
  • So fantastic, can you please delete this question? Now I'm being downvoted for a question I deleted regarding a meta I have no problem with. – MetaChrome Dec 5 '13 at 18:25
  • 5
    Questions with useful answers shouldn't be deleted in general, and meta is different: downvotes don't cost you rep, and what they mean is different on meta - they generally mean "I disagree" which is what my downvote reflects - I disagree that SF's voting/closing is any different than any other SE site. – Ward Dec 5 '13 at 18:29
  • @voretaq7, Do you find no people that are both sysadmin cum developer? – Pacerier May 18 '15 at 17:25
  • @Pacerier No. To be frank you cannot be both: They are two fundamentally different modes of operation. Developers create new and awesome features. Sysadmins keep those features functional in a production environment, integrating them with all the other new and awesome features from all the other developers and ensuring their interactions don't create security or stability problems. Developers and sysadmins should be interacting regularly, but they are distinctly different practice areas, which is why they are separate sites in the Stack Exchange network. – voretaq7 May 20 '15 at 21:09
  • @voretaq7, However, in the first place, the programmer should be responsible that their code don't create security or stability problems when interfaced with other code. Writing crashy code or insecure code is a no-no. – Pacerier May 24 '15 at 13:38
  • @Pacerier "Writing crashy code or insecure code is a no-no" is s a very idealistic and oversimplified view of the world. If it were remotely true there would be no security holes and the Blue Screen of Death wouldn't exist. Developers are human. Humans make mistakes, take short-cuts, and generally screw up whenever the opportunity presents itself. Having different humans (sysadmins) acting as a counterbalance is an imperfect solution, but a vast improvement on one person trying to wear two hats. – voretaq7 May 26 '15 at 21:17
  • @voretaq7, Hmm, on the other hand, there's alot of people with skillset in both areas. – Pacerier Jul 2 '15 at 16:20
8

I'm probably one of the guilty parties here, as I had a lengthy exchange with MetaChrome in the comments of this question, before voting to close it. His/her question seemed to have a middle part (2) that highlighted an interaction between ACLs and default UNIX modes that interested me, but I couldn't get a handle on it without seeing a concrete example. I asked repeatedly for such an example, only be told repeatedly that the question was complete as it was and that no example was needed.

That said, I don't think I can do better than to quote Chris S's comment further down in the exchange:

We frequently find that users, who don't know what they're doing, ask "How do I ABC?", but don't actually want to do ABC, they want XYZ, but it's not what they ask for because they don't know. Hence MadHatter's line of questions.

It didn't help that MetaChrome originally made a simple but major error in his/her question (confusing execute and write permissions). Whilst (s)he immediately saw the error when it was pointed out, the error plus the refusal to provide a concrete example lit up my bozo detectors, and at that point I stopped trying to help and started voting to close.

I should add that although MetaChrome has been very obviously frustrated by the community's response, (s)he has been extremely courteous once (s)he understood where the SF community was coming from, and why it behaves as it does.

So to some extent I don't think this question needs an answer, as it looks like all parties are happy and understand each other. But, like voretaq7 above, I think it's important that these things get resolved openly, and that everyone understands each other as much as possible, so I thought a little context might be helpful.

As an aside, MetaChrome, I still think the question I commented on above has interesting and good elements. I'd got as far as establishing that Ubuntu 12.04.2 uses a 3.5.0 kernel (wiki.ubuntu.com doesn't yet admit of a 12.04.3), downloaded the source for the vanilla 3.5.0 kernel and started poring through the file system code. I've got as far as suspecting a variable current->fs->umask which appears to be the base permissions against which umask is, well, masked, but I can't find out where it comes from. I'm no programmer, so it's slow going, but the question's interesting.

  • 2
    I propose we take a page from LambdaMOO and adopt the Spivak gender neutral pronouns (e laughs, I hugged em, eir heart warmed, that is eirs, e loves emself) :-) – voretaq7 Dec 5 '13 at 17:56
  • @voretaq7 Better than the hermaphrodite pronouns that emerged from the furry MUCK community. – Andrew B Dec 5 '13 at 19:10
  • 3
    I know, you're right, I'm butchering the Queen's English. I hate it. But sysadmin is such a male-dominated field already, I hate even more the thought that someone female wanders in, and gets alienated for no better reason than being addressed as male. That's just rude. I guess having a wife and two sisters heavily into IT leaves its marks. – MadHatter Dec 5 '13 at 19:52
8

Well, this turned into a novel (as usual).
The TL;DR version is that I agree with some of your observations, but differ on why we're seeing what we're seeing.


Is the Server Fault community on a "higher horse" than most other sites about closing questions?

Yes. For two reasons.

The first reason is that, to be blunt, most of the questions we're getting these days are awful. We're talking the kind of things that if I had junior admins asking them I would respond with my new favorite YouTube clip.
We are doing the asker (and the sysadmin community) no favors by becoming a mechanical turk front-end to LMGTFY. There is a running joke about programmers whose only skill is cutting and pasting from Stack Overflow, and I for one don't want that kind of culture in the sysadmin community. We're one of the few professions where it's expected that people can still think on their feet (it's basically us and ER doctors).

The second is that closure of a marginal question is not a death sentence. It takes the question out of the public eye so it can be improved into a good, answerable question.


I have had multiple questions closed because the mods apparently:

  1. Don't understand that the question is specifying details implicitly with regards to what it is asking.
  2. Ie, that the question is specific to a particular instance of failure and a person who knows the answer would easily understand the context of the question.

Well, #1 usually means you didn't specify sufficient detail (or we couldn't pick it out from what you did specify).
Detail should almost never be implied - any good lawyer will tell you that implied agreements are usually worthless and it's only what's explicitly specified that counts.
This is really easy to fix - explicitly give us the relevant details. We reopen a lot of questions this way.

#2 is more of a gray area. If you can improve the question to provide the context do so (context should not be implied either - remember none of us are in your environment). If you think the context is there it's possible the closure is incorrect - flag the question to be reopened or open a meta discussion.
We (both the mods and the community at large) screw up and close things that should stay open all the time. We're human - screwing up is what we do and that's why we have a reopen button.


It is quite vexxing to be patronized because the mods of this particular exchange seem to think that because they do not immediately understand the context of the question with regards to a particular configuration (or that a beginner should or would), starting from the very basics, that the question does not provide enough context or does not exhibit all of the troubleshooting required to get to that point.

We don't expect people to come to us with the answer prepackaged so we can cut and paste it back to them. It's nice when they do, but we don't expect it.
What we do expect (IMHO reasonably) is that some troubleshooting and/or research effort will be put forth and that the problem and information already uncovered will be laid out clearly in an understandable way so we can process it and start generating useful responses.

There's a certain amount of hand-holding we can (and will) do in comments or chat, but Server Fault can't be the sysadmin, and Q&A isn't the best mentoring workflow. We cater to the Exceptional.


Please stop enforcing arbitrary and excessive requirements to provide an introduction to the question and specify the entire troubleshooting efforts.

Like I said above we don't expect people to pre-package the answers for us - just a good, honest, professional effort at gathering details and telling us what's already been done so we don't spend days trying to reinvent the same square wheels that were tried and rejected already.

A question that basically boils down to "My thing is broken. How can I fix it?" isn't really good. "My thing is broken. I looked inside and the widget has come off its sprocket. What's the best way to fix that?" is great.
The spectrum between those two extremes will always be a little arbitrary, until the day the last sysadmin succeeds in our career-long task to automate ourselves out of a job.


Stop closing questions and wait for a person to come along that knows the answer or wait until a bounty is available or wait until the questioner replies with an answer.

Jeff Atwood (of blessed founding memory) was a big subscriber to the Broken Windows Theory.
While I'm not as big a fan of that model as some others I think it applies to this statement -- If a question has problems it needs to be fixed, and placing it On Hold allows that to happen. Ignoring it and hoping it gets some answers just encourages more questions with the same inherent problems.
We try to be gentle and explain what needs to be done to improve questions, but we also try not to leave broken windows.

In the same vein:

It is very vexxing to spend the time writing a question and then getting an inquisition from mods who are not intimately aware of the configuration implying that you have not provided enough information. It is a waste of time and the purpose of the question is await assistance as a convenience.

Well I sort-of agree here. It's vexing to be told "We can't answer this with what you're telling us." To be fair it's ALSO vexing to be asked a question with huge swaths of detail missing and be expected to come up with an answer.

In my mind the purpose of a question is always to get an answer (one or more of them, ideally useful ones). If people are giving "an inquisition" in the comments it's because they feel we need that information in order to give useful answers.

  • 1
    I have to admit I haven't read your whole answer before commenting, but your first response is wrong: ServerFault is NOT faster/harsher about closing questions. Every single meta-site I follow (especially meta.SO) is full of complaints about "you guys are too mean, you close questions to fast." SF is no different. – Ward Dec 5 '13 at 18:06
  • 1
    @Ward Without divulging numbers, the last time I prodded the SE folks for data SF's closure rate (% questions closed) is higher than e.g. Stack Overflow's by a statistically significant margin. (Now as to whether or not that makes us "mean" or just particularly vigilant at weeding our walled garden I really can't say. I don't think we're particularly "mean" with regard to closures, but the ♦ next to my name means I'm by definition a biased insider...) – voretaq7 Dec 5 '13 at 18:49
  • 1
    I recall we were closing huge #s of migrated questions, I don't know if us rabble were told about overall close rates. In any case, I think the harsh "environment" that the OP commented on is more a matter of perception. – Ward Dec 5 '13 at 18:55
  • @voretaq7, What is that video about? – Pacerier May 18 '15 at 17:32
  • @Pacerier "Google It!" (literally what's in the clip) – voretaq7 May 20 '15 at 21:07
2

It is my experience that the environment of serverfault is unlike any other site of the stackexchange network.

It is my experience that that is not the case. I see questions almost every day on meta.Stackoverflow saying "people on Stackoverflow are too quick to close questions, my question is fine, they just didn't understand it." I see questions like that on meta.SF, even sometimes on meta.Christianity.

It's intrinsic to the way all SE sites work that if 5 people think a question is bad it gets labelled "on hold" for a few days and then closed. But on hold/closed questions can still be edited and improved, which will give them a chance to be re-opened.

You must log in to answer this question.

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