*automates your job*

Docker is manageable but kubernetes is way too hard to learn and understand

How is devops software automating my job away? It's just great stuff for creating deterministic containers and having them be able to run everywhere. Unless you're a tech desk support fag this shouldn't be a problem.

Have you ever used Docker or Kubernetes?

I really don't get the point of docker. If you want to develop for a specific platform then port it to that platform, simple as.

Attached: 1535101988135.jpg (1024x1004, 60K)

Umm sweetie it's 2019...

my job's safe until they figure out how to drive thru the worst snowstorm while navigating from A to B and avoid getting stuck on tiny-ass overgrown/oversnowed dirt roads and avoid bumps in the road and carry elderly people up/down stairs and find the right place at the hospital and the elderly home.

which wont happen until they make the humans follow a proper work and information flow to feed it so absolutely fucking nothing goes wrong (which is almost fucking daily).

> and having them be able to run everywhere
I remember that one time when one of our dev leads decided to switch to Docker
> LOL I switched my entire dev team to Macs!
> You don't understand, unlike LXC w/Ansible, Docker can run anywhere!
> I mean it runs on MacOS too and yields the same result
> Presumably
> I was visibly infuriated during the announcement because that manlet pitches it to us sysadmins like it's a second coming of Christ and we should immediately accept the superiour container technology while we were running containers with OpenVZ since, like, 2008
> Demo day, some big G-men are ready to see the future
> Docker containers can not run on the latest Debian (4.9 kernel, mind you)
> Wat do
> Ended up forwarding web ports to his LGBT typewriter to demonstrate the software
Aside from the questionable practice of forcing subordinates to buy hardware with their own money, turns out Docker containers can just not run anywhere for whatever reason. Test your shit, don't fall for hype like macfaggots do.

You are aware that companies are now offering training just for docker security? It is beyond of a mess.

it's some kind of virtual machine that uses less resources and gives more work to the developers
some big companies like google are pushing it on developers to make money out of their cloud servers

Docker will bundle your app code and dependencies into a virtual machine like system that will share your OS's kernel. This saves space and causes fast boot ups. Through docker, you don't have to worry about dependencies screwing you over, and mix matching software versions.

Kubernetes will manage docker for you by restarting containers when they fail and by spinning new ones up if the load demands it.

Needless to say, they are both extremely complicated and verbose for no good reason, and are really hard to get set up with.

Only if you're a brainlet

I can see the utility of Docker but I don't fucking understand Kubernetes. It seems like a giant spaghetti fuckfest of container garbage which requires just as many people to manage. And it's complex as hell.

Do people really find kubernetes hard? I learned it in a couple of weeks and migrated all of our company's workload to it. Wasn't hard at all

That's good, so nothing broke yet? That means you don't know its weak points right now.

It's a spaghetti hack and only worth it if you run many hundred VMs, maybe not even that now that you can do most things with docker-compose (and swarm etc.)

what the fuck job do you have if docker can do everything you do?

I had some issues, but that was 2 years ago, I have since learned all the weak points and pitfalls and fixed them all. Works like a charm.

Do you not upgrade? Do you not run clusters, maybe even geo-distributed clusters?

My setup is relatively simple, cluster of 10 to 20 autoscaling EC2 instances, all in one geographic region. I upgrade to latest version of the kubernetes twice a year.

Are there really credible security concerns if it's used for back rail?

Welcome to the cloud, baby. Jobs are changing, you just have to adapt.
It's honestly not hard for any Linux Sysadmin that knows his shell. Windows admins can start to cry, though.

OK, thank you, and what is your software? RDBs, graph DBs, computational services with in-house software, etc?

How do I learn/get a job with it? There's not many books around about something this new

>haha we're going to meme-automate all your jerbs away, stupid plebs
>OH NO NO NO NO mommy why do I have such a big workload now

Attached: 1549887882717.gif (607x609, 821K)

Do you know AWS or GCP ? If yes, look the doc and start building and destroying things, automate it and do it again with better availability and monitoring.

We run some web apps, backed by background workers and microservices, all dbs are SaaS'd by Amazon, so I don't have to deal with any persistent storage on the cluster. Come to think of it, I guess k8s really can get complex depending on the workload. I got it relatively simple. But im not an admin, my main job is development.

Anyone up for sharing a common workflow that they use day to day with containers? Implying you're using K8s or not.
I just throw up container and let them do their thing; is there any more to management of them besides watching for the security bulletins and intervening on their behalf?

K8S is easy to run - the only pain the control plane - etcd. Everything else is stateless.
You requre more Nodes - you only need some container runtime runc (aka Docker) and kubelet. There are Linux distributions which come with it. That's it. Everything else can be ran by K8S itself - the Container Network Interface (CNI) and all system components are easily run as DaemonSets.
We run over 700 clusters at the moment with over 5000 nodes total and about 100k total containers. And we do that with 5-man team.

The age of Pajeet is over.

> We run over 700 clusters at the moment with over 5000 nodes total and about 100k total containers. And we do that with 5-man team.

Based

You only care about security bulletins for the Nodes which run on.

For the actual containers and images - the less you have in - the better. If you are building a binary it would be perfect as you don't require anything in except for CA certs.

A wild CVE appears - roll new base image and update everything.

Docker is great. Go learn something new grandpa

Docker is pretty sweet, akshually. Once you figure out how to use cloud services and auto-scaling, a lowly software eng can learn to deploy fairly scalable web software with very little money. You can cut the autistic IT cunts right out of the equation and deliver shit on your own.

I don't know what you NEET memesters think but that has been the reality for me on the ground floor.

Attached: yes (3).jpg (400x386, 25K)

How do autoscale?

e.g.
>aws.amazon.com/autoscaling/faqs/

>over 700 clusters
>100k containers
How many requests do you serve per second? People will build the goofiest Rube Goldberg machines with containers in order to be cool and put it on their resume, meanwhile they're serving like 100 requests a minute that could be handled by a 1U with a hot spare

Ex Sysadmin now Sysop, it's great. I no longer have to deal with cisco shit or windows servers, and I get more money for doing a more interesting job

Rundown plz

To be frank - I've no idea. We just manage the clusters and we get some aggregated monitoring data in. What is deployed in them is up to individual teams and their respected services / products.

As you mentioned - there are plenty of legacy software running on them. One of the teams runs some messaging with 1 or 2 million qps running .
The rest use K8S because they can easily run their crap on and on different cloud providers.
Some use Istio for service mesh and traffic management.

Is there anything else then other than docker/kube stuff? Got a feeling than this thing also needs the meta stuff - writing documentation, talking to boomer managers etc

kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale/

You can also autoscale with some custom metrics - request / sec or number of items in queue:

kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale-walkthrough/

do you use some k8s management service or you set it up yourself?

>We run over 700 clusters at the moment with over 5000 nodes total and about 100k total containers. And we do that with 5-man team.
Based employed white person. I wonder what you're doing here, though.

this. what frequently gets lost in all this talk about tech is the stuff you can't just read a manual for. anyone can RTFM and use google. hit the plebs with that meta.

I'm working on taking my RHCSA next month, and then the RHCE shortly after. I feel like I should pick up K8s to be useful for a prospective job, but I have no idea if many companies, besides faggoty SF web-facing startups, are actually really making much use of containerization?

Attached: dog_wtf.jpg (552x528, 65K)

you just apply to McDonalds, you will be fine

I'm not looking to take over your job when you get promoted to shift manager. Looking for legit career advice.

Attached: alex_jones_hurr.png (986x797, 869K)

RHCE and all that uneducated tier certs are irrelevant in 2019, the underlying OS in a dockerized cluster is totally irrelevant, a minimal cluster based OS like CoreOS can start a million cloud VMs with a minimal yaml config file that only contains k8s, etcd and docker binaries, everything else is managed by k8s itself

>he doesn't realize coders will be among the first to go since AI lives in machine code.

Attached: 1543349071802.jpg (612x612, 68K)

>get job
>automate it
>pretend to do the work
>shitpost on 4chin all day
>live the comfy life

Attached: 1553052891163.jpg (184x184, 10K)

>I'm not looking to take over your job when you get promoted to shift manager. Looking for legit career advice.
>on Jow Forums

lmao, yea get that career started with useless redhat certs, good for you man, you'll show those faggoty SF web-facing startups. whatever the fuck that is supposed to even mean

And who manages the underlying OS and infrastructure where dockerization happens? Oh shi-

Attached: hurr.jpg (403x395, 19K)

They figured that one out a while ago. It's why STEM jobs are full of tedious micromanagement.

you're beyond retarded

It unnecessarily complicated, sure, anyone can just COPY files around, RUN some install commands and EXPOSE a couple of ports, but it's when you need to provision when things get thick

Being wrong sucks, eh?

Attached: Trump_smug.jpg (600x610, 67K)

I got a book on it from O'Reilly. Should eventually read it, since I want into DevOps. Then again, something completely different might be trendy when I'm at that point.

>Docker
Imagine makefiles that build virtual machines, and if you run a bunch of the virtual machines at once, they'll share resources automatically and do fancy load balancing things
>Kubernetes
No fucking clue. Frankly, the only things I know are Docker and GitLab CI, since that's what I use for my personal stuff.

containers are not virtual machines, stop that meme and learn how linux namespaces, cgroups, seccomp, bridges, etc... work

that's why I said "imagine," retard. It's a contrived description that's easy to understand.

Docker is shit. Use Podman
developers.redhat.com/blog/2019/02/21/podman-and-buildah-for-docker-users/

>redhat
I wouldn't trust the software quality of a company that maintains yum.

Did you even read the article you dumbfuck? The fact alone that Podman doesn't need root permissions should be a reason for anyone to switch

do you know that kubernetes made docker itself irrelevant? nobody uses docker actually docker by itself, you need a CRI to be contacted by kubelet whenever you create pods, k8s now can just bypass docker and use something like CRI-O

docker as a company actually is in a weird place, it's a universal product but they can't just win the business game with it

Where do you work? What can you tell us about your job?

calm down, lennart. tools like docker and kubernetes are made useful by industry adoption and widespread support.
This isn't an issue that affects the quality of my codebase. It only minimally affects my workflow as a programmer. If I'm running containers through GitLab CI, then I'm not manually dealing with root access either. All I need docker for is running tests, it's not like there's any benefit to the final product from switching. Why should I care about podman?

stop embarrassing yourself, retard

so.... Docker or Kubernetes?

Not exclusive fag

what does this mean? Are you saying they do different things? Why are they always compared then?

I want to invest my time into learning one of these, not both (unless they serve different functions)

'ate docker
'ate kubernetes
'ate devops
'ate agile
'ate startups
'ate frameworks
'ate oop
'ate github
'ate nosql
'ate typescript
'ate python

Love JQuery
Love me microsoft
Love visual studio
Simple as

docker & kubernetes are greatest things ever happened in devops.

differences between them? everything online reads like they're direct competitors doing pushing different software for the same solution

> everything online reads like they're direct competitors
They are, however it's incorrect to compare Docker and K8s. Docker and rkt, Docker Swarm and K8s - now these are really can be compared.

what does this mean?

can you please explain the differences between all of this? everything I've googled still reads like docker and k8s are directly comparable.

I work for a bank, and I've moved a lot of our workload to containers on-prem. We run our apps in containers on Docker 'host' vms, and we have those host vms spread out across multiple physical VMware hosts. The containers are spread out in such a way that we can power down a physical host and not have any service impact. Software updates for our workload are trivial, too. I update a line in a Dockerfile and commit it to source control, then our CI server picks it up, builds a new image, and deploys it. This happens in seconds and involves no disruption in the application. I can upgrade tomcat live in production just by changing a version number.

Dude kops handles this for you

kubernetes "manages" your docker (or another docker like service) nodes.
docker-swarm competes with kubernetes.

good thing it won't happen before the next century

> can you please explain the differences between all of this
Nope, not your personal Google. Difference between K8s and Docker is that K8s is an orchestration system, it can start Docker containers and some other container types, while Docker itself is a container engine, which starts applications in cgroups.

> kops

Attached: 413.png (1277x1149, 39K)

>Aside from the questionable practice of forcing subordinates to buy hardware with their own money
just slap that dumb faggot

Cool shit. Do you guys prefer Docker, or are you stuck with it at this point? Would you switch to another container platform if you wanted?

Kubernetes is most often used to handle Docker images. They're not comparable, they're complementary.

>not your personal google
>proceeds to answer my question exactly as I'd wanted to begin with

Attached: proxy.duckduckgo.com.gif (391x377, 3.08M)

thanks for the explanations brothers

>think docker and kubernetes is great
>my entire software stack is .NET based (not .NET Core)
Fuck.

On the contrary, it makes my job easier. Let's me focus on more pressing things.

It's not a long answer, for which I don't have any time. If you wanted exactly that, good for you.

Who's gonna manage your AD and Exchange server?