New releases

[electrical]

Hi,

We use Shinken extensively in our infrastructure including some of the plugins ( livestatus api for example ) I noticed that a lot of the components ( especially the plugins ) haven't had any releases or commits for quite a long time.

Is shinken still actively being maintained and is any release planned or should we look for an alternative solution?

[olivierHa]

Hello,

Shinken is a framework, so :

"Core/Framework" is maintained (see memory enhancement in June) , and there will be new releases and new features (like maintenance command).

Regarding plugins/addons, it is up to the maintainer, so it is hard to say.

PRs and help are always welcomed anyways :)

[naparuba]

EDIT:

[Ceci était un message demandant à la société Centreon de corriger une de leur publication au sujet des projets Shinken & opsbro, ce qu'ils vont faire. Sujet clôt]

Original post:

Hi,

Thanks @olivierHa for your answer and dedicated work by the way _(/..)/ \(..\)_

Here is the full past history explanation:

• Thanks to @olivierHa and @geektophe the code did have patchs and enhancements during my (fucking long) off time (more than 1 year)
• I did work on both Shinken Enterprise (during the day) & my spare time on https://github.com/naparuba/opsbro
• I did learn a LOT from the Enterprise experience, especially on the user experience:
  • even with IT tools/frameworks, and Shinken framework was very bad on this side from the beginning
  • because I did refuse to make choice on all the Nagios possibilities and did prefer a 100% compatibility. It was an error I think (╯°□°）╯︵ ┻━┻
  • and because I didn't think it was important 8 years ago (╯︵╰,)
• I think more than ever that monitoring as a standalone tool/project is a dead end. You MUST include the monitoring in your full IT process/workflow. Thanks to modules, Shinken is designed for this ᕕ(ᐛ)ᕗ
• I did think that just backport some part of Enterprise into the framework (aka the 3.0 dev branch in git) will be enough and painless.
  • It wasn't
  • Nowadays users want a full ergonomic experience, even for a framework. It your tool is over-powerful but need to read 500 doc pages to understand, you doing it wrong (╥﹏╥)
  • Pickup some part from a full solution into a framework is just too long! Have 2 different code base do not allow to have painless (and test secure!) backports

A quick focus on "But why OpsBro?"

Shinken is a freaking powerful tool for monitoring experts. But you MUST be an expert to setup it. And it's OK if you have expert needs (and if you have a monitoring expert in your team...).

But nowadays, new teams don't have lot of experts. They have Ops that do lot of different things (and it's harder to be good at lot of things that to be an expert of only one specific subject). That's because the business (and a bit of marketing) is pushing them for this. You can't avoid it, it's just the way it evolve.

So I think that there is a problem in some IT parts:

• we have some great experts tools, in several topics:
  • service discovery : consul, etcd, zookeeper
  • automatic detection : like OCS/Fusion inventory
  • monitoring : shinken, nagios, icinga, Zabbix, and other 1000 more
  • metrology : rrd, graphite, influx, prometheus, and 100 more
  • configuration automation : puppet/chef/ansible
  • system compliance : the great rudder :)
• all theses tools are (very) powerful and can just SOLVE the topic they are. If you have a metrology issue, influx will solve it. Even for very advance use cases. Because they are focus on this. Shinken is for the monitoring topic. It did solve it.
• of course all of theses topics are evolving and new uses cases are spawning, and so such tools still evolve and get better, it's an endless run. Each project have it's own strength and flaw, and having choice in lot of projects is better for the user

But there a no "just enough" tool that try to fix "common use cases" on all theses 6 topics.

You guess it, it's what OpsBro is \(^-^)/

OpsBro do not solve all service discovery use cases, OpsBro do not solve all monitoring use cases too, OpsBro do not solve all metrology uses cases.

OpsBro will never be THE and only ONE IT tool. Because no tool can solve ALL theses 6 topics, never. It's just too much.

But Bro is done to help 80% of the users in theses 6 topics. It's done to link with expert tools of theses topics if the user need more in this specific topic.

And here is what I strongly believe: Big one put aside (Google, Netflix, ...) no one need experts tools on all theses 6 topics. They just need "just enough" in all the 6 topics, and maybe 1 or 2 link with an expert tool (like export local node checks into a Shinken to have bp_rule for example, or local node metrics into a graphite for Cluster wide average computation).

I did launch Shinken 8 years ago (already 8 years, time is just too fast...) to solve my own monitoring problems. But now I face lot of small It issues, I spawn new testing nodes, I need to install tools/chekcs on them, automatically add them into a HAProxy, etc etc.

I do not have big needs. I did try Puppet, chef, Ansible, Consul, and such tools. But guess what?

==> They are just TOO much for my needs. I don't want to have a full puppet setup just to have my ssh keys deployed every where (and remove when one of my employee leave).

That why I did start OpsBro (was named Kunai before). As it was shiny and breakable until now, it's where I did try all my UX-fu by putting the CLI as far as I can. And believe me, OpsBro is very far in CLI experience ^^ (Things I can't do so easily in Shinken because here I can't break a CLI call without add isues to the users that need to edit their scripts and production setup etc etc, that's why I did first try in OpsBro before get back all of this into Shinken when it will be rock solid production stable ^^).

Is OpsBro like a monitoring agent? Yes, it's like a monitoring agent, but it's more than just a monitoring agent. Because I think now just be a monitoring agent is not enough. We are in 2017, no more in 1999 when NRPE was started.

That's why I did put all my spare time into it, because I think it's the best I can do to help the most Ops I can in my small spare time (have 3 children is a good way to exponentially reduce your spare time ^^).

What's next so?

Ok Ok, we get the point about the past. (Other Ops project) + (my bad behavior of working alone without talk to others about what I'm doing) ==> a big mess in the Shinken framework project.

I don't think even in getting back in time I can fix all of these, so let's focus on the future:

• First, and maybe the most important part: I plan to release the major part of the Shinken Enterprise into Open Source in the end of the year
  • It won't be FULLy release
    • Believe you can have a company with only a Open Source version in the IT/Ops world is so fucking wrong
  • It will overwrite the current project (after we did merge enhancements from the framework into the Enterprise, because lot of recent enhancement are not in Enterprise)
  • BUT things will be deprecated. Thinks like triggers for example where a freaking BAD ideas. Some advanced parameters are just too smoky for users to understand. so we will deprecate them
  • Some people won't be happy, I know, and I accept it. I think it will be a very small minority.
  • I think user experience must prevail over powerful now
  • If a feature is not usable and understand by a vast majority of the users, won't be OK
• Nagios 100% compatibility won't be a goal. It was a must have 8 years ago. But now I know that 90% of the Nagios/Shinken users are only using a very small subsets of the features. We must focus on this.
• Enteprise & framework won't be de-connected any more
• Some part of Shinken will be ready to accept a OpsBro connection, but I do NOT plan to make OpsBro usage mandatory for Shinken (maybe just the OpsBro lib to share some common code it's it's easier, but NOT launch the OpsBro daemon and have all its files and really use it)

Let's talk about it :)

I'm open to exchange about all of this, and I think in fact the more public it will be, the better it will be (for the project and for me ^^).

I know that some folks will pop out and explain all of this plan is bad, etc etc: But I can motive/assert all the points I did list before:

The Ops habits and needs did change a lot since the last 8 years, let's have a plan to foresee the next 8 years

IMPORTANT NOTE: some people (maliciously) used this post to say my customers that I'll quit my company (Shinken Solutions) and launch a new one based on OpsBro in order to break the incoming deals. This is just wrong. Bro wil just be included in the Enterprise suite but only when it will be production proof.

Ps: I have more than 300Gifs in stock, so if you want to only exchange ideas with gif I'm ready ^^

[Bierchermuesli]

thanks @naparuba for your statement.

I think some people are a bit worried because the project looks a bit unmaintained from the outside world...

for example I suggest to better shut down forum.shinken-monitoring.org instead of collecting spam. Let me know if I can support the project in a topic like this.

btw: why not join monitoring-portal.org ? (long story?)

[naparuba]

Thanks.

Yes that's why the statement was here :)

For monitoring portal it's a freaking good idea! In fact the shinken forum was start when it was still nagios-portal, so it was difficult to set there, and then I never think again about it, but it's a very good idea in fact. I'll look at how manage the transition (archive this one and redirect every new post to there). But it can solve the bogus forum issue

Note: I already try to solve it lot of time with lot of re-captcha plugins and co, but in the end spams always wins :(

[maethor]

Hi @naparuba

"Things will be depreciated because user experience must prevail over powerful." You are completely right!

With that said, I encourage you to list everything that will be removed and to publish this list before Shinken 3 is released, to get advice from the community, or even to prepare them to find other workarounds.

Also, @electrical I would like to add that the Shinken WebUI, which I think is an important part of Shinken value, is still actively maintained.

Thanks.

[naparuba]

@maethor Yes I'll post the deprecated part before doing it, so we can exchange about it. But I think won't be a major problems, as it will be only part that are not used anyway ^^

[maethor]

I'm actually using triggers (It's not a problem, I must admit I can find others and easier ways to do it). So excuse me if I don't trust you on what you think is not used anyway :)

[naparuba]

I did post an edit to explain that I'll NOT quit my own company to build a new one based on OpsBro.

I though it was obvious, but seem that some of my french market opponent are prone to do dirty things to get my company down ^^

Good try guys, good try:

[dacou]

Hello, "First, and maybe the most important part: I plan to release the major part of the Shinken Enterprise into Open Source in the end of the year " Any news about this ?

[JSFonta]

Hey @naparuba , You talk about a new Shinken's version from Shinken-Solution. One year later, can you provide us more details please (Realease date/period) ? Also, thanks for all details and those wonderful gifs ;)