Looking for early history of Continuous Delivery

Humble and Farley's book from 2010 titled Continuous Delivery may have popularized the term, but does anyone know the early history of this term? At least, how did the concept evolve? Does it mirror the evolution of DevOps?

See https://devopedia.org/continuous-delivery for a brief and incomplete history.

https://redd.it/orbr79
@r_devops

What do YOU do with Python?

Or other script languages? I'm curious and would like to hear some real-world examples, or even better, see them if you can share.

https://redd.it/orctqs
@r_devops

Plan your work with Azure DevOps Board

https://youtu.be/ozFfmLeH-eg

https://redd.it/oreccs
@r_devops

DevOps interview question, would like some insight!

Hey guys, long time lurker here and wanting to go over a question I was asked in an interview and see how I could have answered it better. Been doing dev work for sometime and this was my first DevOps interview and most of my knowledge of the field has been self taught and hobby projects.

I think a lot of it came down to nerves and getting in my own head but I want to use this as an opportunity to research and learn more to be better prepared for next time.

The scenario I was asked in the interview was:

>The team is ready to deploy an app, they have a docker container that listens to http requests on a specific port. They want to be able deploy the app to multiple servers and environments, and after it has been provisioned have the docker container run with the port exposed. The app is in an alpha state so it is prone to crashes.
>
>Describe how you would provision the server.

My answer, in short was to use Terraform to spin up an instance for Azure,GCP,AWS (offer a cloud agnostic approach), and from there utilize Ansible to set up the server and deploy the docker container with the --expose [specific port\] and --restart=always.

Was this the correct approach to answering a question like this or should I have focused more on detailing out the Change Management tool as apposed to calling out something like Terraform?

How would you guys go about answering or heck even performing something like this?

https://redd.it/or7p7d
@r_devops

Docker Images by Proyect or by Type

Hi, how are you? My question is whether I should build docker images by project or by type.

Example: An image for the front-end projects or an image for each specific front-end project.

It could also be both and that one is built from the other.

What do you think? Greetings.

https://redd.it/orgt4l
@r_devops

Help with take home assignment

I've been given a take home assignment (I know I know) around writing a script to do a dB upgrade.

----STARTS_HERE----

Please go through the described scenario and write a script, in one of the below languages, implementing a fix to the issue below. 
 
For the development of the scripts you have 4 hours and are allowed to use Google and any other material as long as the work submitted was written by you. 

Use Case: 

·   A database upgrade requires the execution of numbered SQL scripts stored in a specified folder, named such as '045.createtable.sql'

        - The scripts may contain any simple SQL statement(s) to any table of your choice, e.g. 'INSERT INTO testTable VALUES("045.createtable.sql");'
 

·   There may be gaps in the SQL file name numbering and there isn't always a . (dot) after the beginning number
 

·   The database upgrade is based on looking up the current version in the database and comparing this number to the numbers in the script names

        - The table where the current db version is stored is called 'versionTable', with a single row for the version, called 'version'
 

·   If the version number from the db matches the highest number from the scripts then nothing is executed


·   All scripts that contain a number higher than the current db version will be executed against the database in numerical order
 

·   In addition, the database version table is updated after the script execution with the executed script's number 

 
·   Your script will be executed automatically via a program, and must satisfy these command line input parameters exactly in order to run:

        - './your-script.your-lang directory-with-sql-scripts username-for-the-db db-host db-name db-password'   

Requirements:
 
·   Supported Languages: Bash, Python3, PHP, Shell, Ruby, Powershell - No other languages will be accepted

·   You will have to use a MySQL 5.7 database    
 
How would you implement this in order to create an automated solution to the above requirements? 


-----ENDS HERE----

What would be the best way to go about this?

I'm not sure what they mean by my script will be executed automatically. Is it better to write the script in python or powershell?


Thanks
 

https://redd.it/orbpcd
@r_devops

CI/CD experts here?

I was cleaning up my Download folder and come across PDF named cicd.pdf. It contains only 2 images

https://www.dropbox.com/s/pceafrhx9ivpue5/cicd1.png?dl=0

https://www.dropbox.com/s/9yx5l2ms72dla63/cicd2.png?dl=0

and absolutely no text/description. I don't consider myself as an expert in CI/CD, that's why I need your feedback. So far I am not sure should I keep this PDF or not. At first glance this could be an ideal pipline, created according all the best practices.

​

1. Build once, deploy on any ENV
2. Swarm to fix master if ppeline fails
3. Static code analysis and security scanning
4. Every commit is potentially shippable artifact.

​

On the other hand, is it right decision to build DEV env from the scratch and kill it in the end of CI stage? Building environment could take time and as far as I know CI stage should not be longer than 15 min (actually the faster the better). But on the other hand instead of building whole env, with all microservices we can deploy service-mock just for integration test purpose. Because it is impossible to test anything having one DEV and 100 microservices/pipelines.

Am I right, this is trunk base development? Because I don't see here -SNAPSHOT java artifacts, so every commit/artifact is RELEASE and could be deployed.

https://redd.it/orrrt4
@r_devops

Sonarqube project links

Hi All,

I have been exploring sonarqube for a while and I have some queries regarding how sonarqube detects a project.

Usually I create a project name in sonarqube which resembles my gitlab/github project and then scan the code repo for code issues. Is there any way that sonarqube can detect the repo info by making use of git info. Or am I doing it in the correct way.

And another way I found to attach the gitlab/github info is by using project links where I was able to add thr info. Is this the only way where we can actual project reference to sonarqube project?

https://redd.it/orud3e
@r_devops

Do DevOps engineers need to be architects?

I guess I've been fortunate in that I've always seemed to land at a place where there were more or less defined boundaries between dev teams, components, and how they talk to one another. This made onboarding relatively straightforward as I knew where to draw a line and say, okay, everything below this line I don't need to know in order to do my job. I don't need to know, for example, how a particular method in a particular service manages state, but I do need to know that the service connects to a Redis cluster, and it will fail to start if that cluster is unavailable.


About a month ago I started a new position where I've thus far been utterly incapable of finding where to draw that line of abstraction. There are \~15 developers working on 60 "microservices" deployed on ECS. I use quotes because what's really going on is one giant sprawling shared code base running as 60 different Python scripts and communicating through at least six different technologies - an object store, some databases, a message queue, a key-value store, and some synchronous REST API calls thrown in for good measure. Multiple services are handling the same data in different formats. Unsurprisingly, the system is a massive resource hog. I wouldn't be surprised if 70% of our compute costs are being spent on serializing and deserializing the same JSON over and over again.


When I sit down with the devs to try to make sense of what's happening, my eyes quickly glaze over while clicking through the code together. When I ask, "why did you decide to implement it this way?" I usually get a shoulder shrug. I'm coming to the conclusion that no one really gave any thought to how to build or operate this thing. Consequently, I feel like I'm grasping at straws trying to impose order or find patterns in a system where there really aren't any. This is extremely discouraging and exhausting.


I'm tasked, among other things, with migrating this system to a more standard orchestrator (they specifically want to migrate to kubernetes), but I feel like the amount of effort required to do so given the current state of affairs would be prohibitive in terms of time and effort and would not yield the desired benefit over the current setup. My superior was receptive to this assessment, but now I need to come up with a plan for what to do instead. This is where I'm stuck.


Obviously I can't tell the devs to drop everything and refactor the whole code base according to some shiny design pattern I found in a book. I'm looking to make friends, not enemies, so I'd like to propose whatever it is I end up proposing we do as a way to improve reliability, scalability, and performance, because so far we are scoring dangerously low on all three. Everybody knows it but nobody seems willing to take charge and lead the way forward.


Should I stop trying to be an architect, put my head down and start banging out YAML?

https://redd.it/orsy3v
@r_devops

Ansible Tutorial | How to write Ansible Playbooks | Ansible Crash Course

If you want to see the full Ansible Playbook Crash Course Tutorial please visit this link:- https://www.youtube.com/playlist?list=PLTCuRW0ikUdPYxoiVlgDnrySA4eCu2d08

https://redd.it/orvdae
@r_devops

Why I get secrets "argocd-initial-admin-secret" not found?

Hi experts,

I have problems in getting argocd to work for me in the vagrant machine.

Here's what I did:

vagrant@localhost:~> kubectl -n argocd get secret argocd-initial-admin-secret -o jsonpath="{.data.password}" | base64 -d
Error from server (NotFound): secrets "argocd-initial-admin-secret" not found
vagrant@localhost:~>

and this is also from the guide :

https://argoproj.github.io/argo-cd/getting\_started/#4-login-using-the-cli

Can someone tell me how to make it work?

​

Tks.

https://redd.it/orwm7o
@r_devops

Is there Webserver (nginx) with built in api for creating config?

I'm from a world of K8s and using things like the nginx-ingress-controller. Where I am currently we don't use k8s, but I would love something similar for doing our testing environments routing.


Does anyone know anything that exists, for example, an Nginx instance with a built in api that I can POST/DELETE to which will create / remove a rule and reload the config. Similar to how the N-I-C will detect an Ingress object and create its respective rules?

https://redd.it/orv8wy
@r_devops

3 ways to make DevOps continuously better

According to Gene Kim, author of Phoenix and Unicorn project - both books on DevOps - teams should focus on "the 3 ways" as a non-stop operational loop. I've done a very high-level personal interpretation of the 3 ways below:

The First Way: Principles of flow \- think of the system as a whole rather than discrete components and as the flow of information and activity throughout

The Second Way: Principles of feedback - create right-to-left feedback loops that listen to internal and external customers to amplify their experiences and shorten the learning cycle

The Third Way: Principles of continuous learning - learn through experimentation and bringing in new ideas that can be implemented for gathering data

How do you put these various ways into practice in your current software engineering situation?

https://redd.it/orwtoj
@r_devops

How do I continue to improve as a solo devops engineer?

I joined a small company a few years ago. The thing that motivated me was that I will be able to build a cloud based system from a ground up. However, now the development really slowed down. We have a niche b2b product that is used by small set of customers. I don't feel that I'm learning much in this job. However, my salary is quite high. I tried applying to other jobs and found that a) we are in the much better state then other companies in terms of devops adoption 2) salary in my current position is quite high and I can't expect much of increase if I join some other company.

Therefore my plan is to spend an hour each day to improve and learn some new skills. Do you think is a good idea? What platforms can you suggest for learning stuff? One of the areas I wanted to improve was k8s. We don't use it and it'd be impractical for us to move to k8s but it seems that the knowledge of it is required in almost all devops jobs these days.

https://redd.it/orxf6f
@r_devops

The late night deploy

I love the company I work for--good pay, good culture, very good benefits, but I'm venting a little to express my frustration. I would say we're marginally profitable, like the owner lives in a house that is a pretty normal house and drives a 2014 volvo station wagon, so it's not like we're killing it on the revenue side. However, all of that aside, we do late night deployments and host our own private cloud/data center, have no blue/green or anything that would permit business hours deployments/updates. The company is 12 years old and despite all of the perks, I'm astonished that they haven't felt the urgency to address these concerns to create something that is closer to a 9-5 schedule for the employees.

While i had some initial enthusiasm, everyone says it's not in the budget and the DEVS are also just used to this because this is how it's always been. Feeling demoralized.

I would consider going elsewhere but this is my first DevOps gig and i'm learning a ton and being well mentored. I doubt I could pass a coding interview elsewhere or stand up a CI/CD scenario with any efficiency. But just frustrated by the late nights and the things that are clearly avoidable if you have your priorities in order.

https://redd.it/orxn6s
@r_devops

I have difficulty being DevOps. Please advise.

Good day my fellow IT professionals, I'm in a bit of an anxiety-ridden pickle and I made a throwaway (?) so I can freely talk.

I'm almost three months in my new job for a Large Tech Company and I'm pretty cool with the company so far. My role is DevOps, and I have quite some difficulty adjusting to it, however.

My previous company is very traditional and has been using outdated practices and technologies (read: Waterfall, Java 7, name it). I am a Computer Science graduate, and my previous work experience is purely database development (I write complex SQL queries every day). In short, I have zero experience in the operations/infrastructure side of things.

I was told by my new boss that I was hired since I am proficient in Java and databases, but my daily tasks these days have very little development and more on supporting users having difficulties on our platform.

I feel like I am not going to perform well being entirely new to the DevOps scene, and I feel I won't be regularized by my 6th month. It doesn't help that the new hire, who started on the same day as I did, is able to do more, since he has experience being operations/support.

Please tell me what should I study more so I can finally adjust to this position. I want this job, but I feel like I'm lost at the moment. Thank you very much.

Notes:

- We aren't using stuff like AWS or Terraform like what I've been seeing in this sub. We use Kubernetes/Docker, and a mix of Grafana, Prometheus, Kong, Kibana, and other tech I have no clue what they're precisely for. Like there's so MANY tools to study about, it overwhelms my ADHD brain.

- I wish it could be as easy as "talk to your supervisor about it" but he's currently in the hospital.

https://redd.it/orxhao
@r_devops

Beginner re-deploy service upon push to master creates docker image artifact

Hi,

I'm quite a beginner. I'm working in a big company and I have a question.

We have internal service that uses HTTP requests.

Anyway, it's behind nginx, inside VM server.

What we would like to do is, when pushing to master, the pipeline already creates docker image, and pushes artifact. But what we also like to do, is when it finished pushing artifact, we want to stop the current service (which is outside the 'CI/CD slave/worker'), remove container, remove image, docker pull image from artifactory, and re-deploy the service, with some parameters.

Is that possible?

Currently, I do this manually (docker stop, docker rm, docker pull, docker run).

https://redd.it/orzhiz
@r_devops

Runbook.md -- Bash executable runbooks in markdown

Runbook automation and literate programming are both not new ideas, but I think the two play well together. So, I got this idea last week, and hacked up a simple tool that I think would be useful to SysAdmins, Ops people, or SREs. Please check it out - https://github.com/kjkuan/Runbook.md

https://redd.it/os3lnd
@r_devops

What is the most mentally challenging stuff that you had to work on ?

I am feeling like having a lack of mental challenge, creating Kubernetes yaml files, troubleshooting jenkins pipelines or writing HCL all day long... It was fun at the beginning but begin to be bored.

I am hoping that somewhere outhere, some people had the possibility to work on something more rewarding.

https://redd.it/os4p07
@r_devops

Unpopular Opinion Terraform HCL is fedding me up

Although there are some alternatives now to HCL, people still advocating for using HCL for it's simplicity and readability.

While in fact it is a real struggle using it at scale, it is hard to template anything in hcl, hard to validate, hard to extend, hard to integrate with other systems, hard for secret management, hard to maintain due to versions change and provider instability, it's hard to make smart changes, hard to use the state versioning, hard to loop through variables, hard to check drifts.

Other alternatives include terraform cdk, and all other templates systems like jsonnet, CUE, etc.

How do you handle of the hard tasks I just defined.

https://redd.it/os3elx
@r_devops

Need help in architecture of a cloud gaming service - Storage

Hi folks,
I am trying to create a cloud gaming service and wanted to understand how can we move games across multiple user sessions/vms easily like GeForce Now or Google Stadia.

For example we have a set of 1000 game virtual disk running on ssds on a SAN.

How can we securely give access to user and also maintaining zero user logs or that means anything the user does with the virtual hard-disk should be immediately deleted as soon as the user leaves the platform.

+

What if multiple users are requesting to play the same game, do we let say copy the virtual game image from original storage to let say Servers Ssd storage and unmount the disk ?
Dont you thats gonna be too much IO and will kill the SSDs in a year by doing let say 2-5TB read/writes per day (avg game size 50gb)

https://redd.it/os7tgl
@r_devops