What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 138 Stack Exchange communities.

In the last month or two I've noticed my questions are becoming increasingly dev ops related and less related to server fault or stackoverflow. Examples include 0MQ compiling issues or problems in TCP.

The field of Dev Ops is ever increasing, and I know teams who have rewritten protocols and entire NGINX modules but their problems aren't necessarily software or hardware related. They can be hybrid problems or caused by a programming error that didn't respect a firewall keep-alive.

I think many users may be confused as to why they see some of these questions and simply say they're off-topic. But now that our economy is cloud-based and SaaS companies do not control the hardware, Dev Ops have high value in the field of software development.

Where do we put these off-topic questions?

Update: example

This question was about 0MQ and probably only someone working in Dev Ops would have a clue. It was multi-faceted and involved Java bindings, Visual Studio, and C compiling

Definitive fool-proof steps for 0MQ / ZeroMQ and Java on Windows 7?

This question was about redis which both requires networking, server and software expertise. It took some time to track the bug to the Scala client itself:

Scala-redis subscribes to * but receives zero messages

share|improve this question
Consider including some links to some examples in your question that were actually closed as off topic. This will give everyone a better perspective of what you're seeing. –  jmort253 Jun 2 '12 at 19:36
Admittedly I deleted another question earlier today as it was about to be closed. I did so since it was too open-ended and could have been a problem in our websocket server, our host's firewall, or a combination of other things –  crockpotveggies Jun 2 '12 at 19:47
I completely agree to the intent of DeLongey's question. I would ask questions such as; "How do I handle deployments with Capistrano, Puppet and Ubuntu/Upstart with continuous integration?", or "What is the best way to bootstrap an EC2 instance into a cluster and handle the synchronization and installation needed to make my own services start properly? - Would you use Cobbler+Puppet+Capistrano+MCollective+ZooKeeper e.g.?" "What is a reliable failover configuration configuration with HAProxy and Keepalived?" "Would you use AMIs or TFTP if you're bootstrapping your devs with vagrant/VirtualBox?" –  Henrik Jul 24 '12 at 11:52
"What is the best way to do rollbacks of hetrogeneous service oriented architectures with continuous integration?" "How do you manage versioning of services w.r.t rollbacks with TeamCity?" "Is there a way to deplace dnsmasq with a quorum-aware service like ZooKeeper?" "Can I replace DBus with ZooKeeper somehow? - both do signals, but DBus is single point of failure while ZooKeeper is quorum based" "How do you analyze your system logs and application logs and how do you integrate them with e.g. Logstash and ElasticSearch?" –  Henrik Jul 24 '12 at 11:56
"How do I set up a continuous integration environment that does a run through my isolated services, in a well-behaved manner - can I automate VirtualBox with Vagrant/Sahara/Puppet in TeamCity and how do I synchronize queue drainage from the message broker?" "What do you use for 'faulty failure detection' in your distributed systems?" "Can I use OAuth to authorize my services across programming languages rather than having a certificate authority handle the individual computers and then having a different layer of security ON TOP OF that, doing OAuth + Claims based security alone?" –  Henrik Jul 24 '12 at 11:59
As you can see, I have a few questions that I would like to have answered. It turns out that these questions don't really arise until you start working with distributed systems, clusters and failover management - as well as all the rest I mentioned above - as a part of your daily job. And it's pretty hard to find a cohesive resource, but imo DevOps deal with all of the above questions, because it facilitates the knowledge exchange between operations/sysadmins and devs that is required to know the answers to the above. –  Henrik Jul 24 '12 at 12:01
"For those of you using nodeless puppet configuration; what master election protocol or logic in puppet are you using to select what nodes get to be master/slaves? E.g. HA-clustering with RabbitMQ - you need to choose one or two disk nodes and a mem-node - how do you select which gets to be which?" –  Henrik Jul 24 '12 at 12:17
stackoverflow.com/questions/9011674/cluster-monitoring - supposedly not a real question - despite that I have answers to it! –  Henrik Jul 24 '12 at 12:31
Just a thought...but "StackOps" or "StackScale" LOL –  crockpotveggies Jul 24 '12 at 22:02
Didn't even know there was such a thing as Dev Ops (we just call that "software engineering"). –  user7116 Sep 27 '12 at 17:00
@Henrik why don't you ask some those very good questions you posted above, in here - area51.stackexchange.com/proposals/66151/devops ? –  guido Oct 9 '14 at 10:22

2 Answers 2

I think there should be a site for it.

share|improve this answer
You're free to do so. –  Time Traveling Bobby Aug 28 '12 at 15:08
there is a proposal for just this –  warren Mar 14 '14 at 14:19

In a way, many of these questions sound to me like a subset of the questions that are on-topic at SuperUser: how to install, set up, and use some apparently poorly-documented software. I think the deciding factor as to whether this would make a good Stack would be whether there is actually a population of people who know the answers, or if the knowledge is spread so thin that most questions would go unanswered.

Other questions though, begin "What is the best way...," "What do you use...," etc, and those are the sort of discussion/polling/shopping questions that are banned at most (if not all) stacks. I'm concerned about those; several of the ones you posted in the comments above don't sound as if they could be rephrased another way.

My remaining concern is about rot. Those questions all sound extremely specific to particular versions of particular software packages used in combinations that would come tumbling down like a house of cards as the various packages evolved. For most of the questions on StackOverflow, the answeres will still be relevant in a year; I somehow don't think this is the case for your proposed site.

But ultimately, the proof is in the pudding, and my concerns may be invalid. Therefore, I'll point out that there is an established process for requesting and initiating a new StackExchange site: Area 51. Head over there and work up a proposal, see how it flies!

share|improve this answer
I think some of your points regarding phrasing and time-relevance are indeed valid. The success of such a site might be in the format, such as emphasis of posting date. A helpful feature to prevent it could be to auto-flag a question and mark it for follow up. "Best practice" changes every day. –  crockpotveggies Jul 29 '12 at 22:11
Many of the questions don't have exact answers, and that can be said about programming as well, which StackOverflow is about. Not everyone finds 'Why is library xyz throwing ErrorABC when I turn the corkscrew?' very interesting - because of exactly how stale those questions will be in a year or two when the library authors have moved on to new bugs or other software is in use. On the contrary, how a quorum is assembled is timeless, because the underlying computer science is. Just because it's a normative question, doesn't mean it can't be answered objectively. –  Henrik Aug 29 '12 at 7:03

You must log in to answer this question.

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