When you're using it as a noun, it's walkthrough - one word.

Can you tellme how/why?

Hey Ravi, thanks for the comment! If I’m understanding your comment correctly, I feel the same way. The productivity boost from having the tool baked into the IDE isn’t enough if the AI model (codewhisperer) isn’t comparable in quality. I’ve found that I still save time by working with ChatGPT and doing the copy-paste dance rather than trying to get codewhisperer to work directly in VSCode. The quality of response is just better with ChatGPT at this point in time for me personally. I think the more interesting comparison would GitHub Copilot and ChatGPT because I think there’s enough there with Copilot that I could feasibly make the switch, but I just haven’t tried it yet.

Hey norandomness, thanks for the comment! As stated in the video, I simply don’t see how AI will be able to replace complex human interactions on an organizational scale in any short timeframe. I’m open to being wrong, but two years is definitely way too short in my opinion. If AI is able to replace DevOps in two years, we will have bigger problems to address in the industry as a whole with mass layoffs of all sorts of engineers, not just DevOps. I simply don’t see it happening. It takes an expert to figure out how to tie all of these systems together and GPT-style AI models are simply generating content of various types, not actually reasoning about a system and knowing how to cause a desired change with no oversight.

​@@JarrettCogginnever mind that Fearmongered people lol they don't know how ML/AI works and what's the limit and platoue. They just spit out fiction based on their nonsensical fiction in their heads.

@@JarrettCoggin +1

One year

​@@JarrettCoggin hey can you tell about data engineer role?

Thanks for the feedback Henderson! Will definitely improve this in future videos.

Thanks for the comment CodeTrix! I spend a bunch of the time in the terminal, and most of the time I see people testing out generating code for Python, C#, Java, etc. in these Code AI tool videos, so I wanted to see how it would work for terminal work as well. Anything else you'd like to see specifically?

Hey Greed! Thanks for your great comment. It's great to hear you found this helpful! I think it's awesome that you've tackled these problems in the past at the fintech startup. It can be a lot of fun and great satisfaction to have a low-friction way to ship code out to various environments, including production. For me, a pipeline is the end-to-end process from commit to live in production for a single component (a component being a front-end UI OR a microservice OR a database). A way to think about it is that every component in a system has a pipeline, but some of the steps in that pipeline happen to be manual. Writing all of the steps out by hand can help you uncover the pieces of automation that need to be implemented (or eliminated) in order to automate the entire pipeline end-to-end. Based on what you described and assuming that you have a DEV, STAGE, and PROD environment setup, my initial thoughts would be that: 1. You'd need to decide how you are hosting the individual components, whether that be on a VM (such as an EC2 instance) or on an orchestration tool like Kubernetes. That will help you decide whether you need a configuration management tool like Ansible at all for the application's components themselves. You may not need the configuration management tool for the application, but you may need it for your infrastructure type components like your Jenkins controller, Jenkins Build Agents, and your Kubernetes nodes. 2. I'd recommend drawing out a diagram of what environments you plan to have, what infrastructure you need, such as how many AWS accounts you will have, CI/CD infrastructure AKA Jenkins, how built artifacts like images end up in various AWS accounts if you have more than one or are they being pulled from one central location cross-account, etc. Going through this process can help you investigate authentication, authorization, and security items that may be a little more involved. At the very least, it will help you ask targeted questions to accomplish specific tasks, such as "How do I pull a container image from an ECR repository in another AWS account?" If you have secrets (which it sounds like you will), how will those be stored and how will you pull them from storage to run your application? 3. Draw out the pipeline process that each of those components needs to go from commit to live in production. Think about the steps needed to have an automated roll-back as well, but that's a later phase of implementation. For me, a pipeline typically has a build phase where you run your compilation step (if needed), linting, SAST, and image building, and when you are done with all of that, you also publish the container image to a repository. Once you have a container image, you need the pipeline to then kick off deployment to DEV, and if you have automated tests, you then run those automated tests. If the tests pass, the pipeline automatically proceeds to STAGE. If they don't, the pipeline stops. You would probably repeat the same process for testing in STAGE then rolling out to PROD. You also get to decide if you have a true "Continuous Deployment" style CI/CD pipeline where there is zero manual steps from the time code is committed to git until it is live and running in PROD. If you do want a manual step in that pipeline, it can definitely be done with Jenkins with a "Manual Input" step. I'd encourage you to write your pipelines with code, if possible. This makes it easier to share pipelines across components as a project grows. 4. How will logging, tracing, and monitoring fit into this? You mentioned Nagios, but will Nagios also work for logs written by application code itself? I can't comment too much on Nagios since I don't have much experience with it. Would definitely be an area to look into. Other options may be Prometheus or a tool like Datadog or Splunk, but Datadog and Splunk are paid commercial tools, so they may not be feasible. If you have any questions, I'm happy to answer them and I hope you don't mind the unsolicited advice.

Thanks Lynn! Good to see you!

Thanks for the feedback, Nikhil! Will improve this in future videos!

@@JarrettCoggin higher contrast would help too

Hey Nick, thanks for the great idea! Will definitely incorporate this as well!

@Krishna Prakash, great question! I have two different answers for this. One answer is about specific skills to target for specific jobs, the other answer is for proving you have those skills. For my first answer, if you are trying to go for a specific type of job, say DevOps Engineer in a mobile-centric organization (think Lyft, Uber, Twitter, Instagram, Tiktok, etc.), then you'll want to identify specific tasks/projects/skills that might be more relevant to that role. You'll have to look through job descriptions at those companies to figure out what problems they might have and what skills they are looking for to know what to target. In one of these mobile-centric companies, you might look into what it takes to build, deploy, and test applications on mobile devices. For example, what does it take to upload an app to the App Store? Are there any specific security concerns you need to be aware of? How do you run automated tests in a reasonable time frame? Are there specific resources you need to do that type of testing? Can you even do all of this in a cloud provider or do you need to host infrastructure on-premise? In the case of working at a Data Science/Artificial Intelligence (AI)/Machine Learning (ML) company, what does it take to train and deploy new models? How do you incorporate a built and versioned model into an application? How do you host the models so they can be used for inference? Does the model need GPUs to run effectively in production? Can the model be deployed as part of a mobile app? All of these questions change the types of skills required and the questions a DevOps Engineer will face at different companies or different teams/orgs within a company. There's a whole bunch of different areas you can get into for identifying skills to build when you start making the hypothetical environment more complex. What if the company is using multiple regions in a cloud provider? What if they are using multiple cloud providers? What if they are hybrid hosting and using a cloud provider in conjunction with on-premise infrastructure? What if they are deploying web apps, mobile apps, desktop apps, and machine learning models? My second answer would be to put together a project showing end-to-end how all of these tools and skills work together. This is really helpful to truly develop the skills in a concrete way and can be a good opportunity to fill in any knowledge gaps one may have For example, if you can put together a project that shows that you know how to deploy a simple web application that has a front-end, a backend API, and a database onto a cloud provider via a CI/CD pipeline (ideally all written as code), then you are in a great spot. For example, can you meet all of the following requirements? - All of your AWS infrastructure was defined and provisioned with Terraform and can be fully set up and torn down by only running a few terraform commands (one to initialize the code, one to apply the infrastructure, and one to destroy the infrastructure). - All of the application code for the frontend, the API, and the database is stored in Git (probably GitHub). - The frontend of the application is hosted on an HTTPS-only endpoint. - The end-to-end CI/CD pipeline is defined entirely as code and is automatically kicked off when a new commit is made to the Git repository. If you are using GitHub as your Git repository host, you can easily add in GitHub Actions as your CI/CD pipeline execution engine. - Your CI/CD pipeline has well-defined and separated build/deploy/test phases. - Your CI/CD deploys to separate non-production and production environments in a completely automated way. - Ideally, for true Continuous Deployment style CI/CD, a code change should require no manual intervention at all to get from initial commit to the git repository all the way out to production. - No authentication credentials are checked into any repo. - If you are hosting your build agents in your AWS account, the build agents should be dynamically provisioned from a pre-built AMI that you have created via Ansible (and probably via Packer). This will slow down your CI/CD pipeline, but it will showcase that you don't need any manual work to tear down and set up known-good build agents. You may need to do some bootstrapping here where you build a portion of your infrastructure by hand until you have automated your build process for the Build Agent AMIs. There's definitely more constraints you can put on this project, but this is definitely enough to get started with. Let me know if you have other questions!

Hey there, I've been thinking about your comment for days now and I can't figure out what you are referring to. What do you mean?

@Liquid Mobius, thanks for the comment! With the way I answered the question, I definitely can understand how it can come across as denial. I'd put myself in the "mild skeptic" / "cautious optimist" area of the fanbase, if you will. I really WANT AI to revolutionize the landscape of Software Engineering, Computer Science, and Information Technology, but I don't quite think we are there yet. Since GPT3 and ChatGPT blew up, I've thought about this question in the context of DevOps Engineering for a while. I've been using those models for a few months now on a variety of tasks (both personally and professionally) and have friends and colleagues that have done so as well. My impression of the current generation of AI is that it's great for getting off of the "blank slate". The answers I've seen it give are 80% of the way there. I've seen it generate large amounts of code, and it gets the majority of it right, but it also misses the nuance of the situation and introduces security issues in just about every answer it has given me. If you prod the AI enough, you can get most of the way there and address some of the problems, but it's never just a plug and play implementation. The code provided ALWAYS needs adjustments and human review. I also think one of the great benefits of AI is the ability to have a smart "rubber duck" to bounce ideas off of. I greatly appreciate the ability to have a conversation with ChatGPT about technical subjects I'm either well-versed in or only loosely knowledgeable about. For the well-versed topics, occasionally ChatGPT will present some angle I haven't thought of, but usually it will present things I already know and sometimes dumps out flat-out wrong information. On the other side, it's a great tool for helping me quickly get up to speed on a topic in a way that I feel like I can do more research, have concrete directions to go in, and that I can ask smart questions, but at the same time, I don't completely trust the information from things like ChatGPT because I know it's not completely accurate. Will these issue be resolved? Most likely. At the same time, there is a lot of what I'll call "context" that AI just can't grasp yet because AI doesn't have access to, isn't integrated with, or can't be combined with (social structures, team dynamics, meat-space type stuff). Tech companies have been working a certain way for decades, and I don't see this changing anytime soon. I think if we are going to see AI address anything on a meaningful scale as one would hope, it would be at the smallest companies with the least resources (startups).

@shrekisntlove7312 - thanks for the question! I was not able to find anything in The Linux Foundation Training and Certification Handbook that indicates there is partial credit/marks for problems in the exam. I would assume the certification exam does not give partial credit. Here's the specific page in the handbook: docs.linuxfoundation.org/tc-docs/certification/lf-handbook2/exam-scoring-and-notification However, the killer.sh mock exams do give partial credit and they show you very specifically what steps were not completed correctly. For example, imagine that you have a question on the mock exam that requires 6 different steps, but you get only 4 of the 6 steps correct, the mock exam will tell you the specific steps you got right, what you got wrong, and what the correct steps are. Let's say I have to set up a deployment, a service, and an ingress in order to publicly expose an application, and I get the deployment and service right, but get the ingress wrong, the killer.sh mock exam will tell me what I should have done for the deployment, service, and ingress steps, including specific commands to run and what the manifest file should look like. It's very helpful in generating targeted practice and figuring out exactly what you should be doing to review!

where was he coding?

@cffoo438, thanks for the great question! Because I already had some familiarity with Kubernetes (created Pods, Deployments, ConfigMaps, etc. already), I've found the A Cloud Guru CKAD prep course the best, although it starts off a little odd by jumping straight to Jobs and CronJobs after building container images. I would recommend this course first IF you already have some Kubernetes familiarity and you are looking for an in-depth refresher with hands-on exercises to help you get prepared for the exam. If you have little or no familiarity with Kubernetes, I'd recommend the Udemy course I mentioned in the video description. It builds your learning more from the ground up with Kubernetes and has hands-on exercises along the way. Even though the Udemy CKAD course is technically 1.5 hours shorter than the A Cloud Guru CKAD course, in my opinion, the Udemy CKAD course is a longer course because it has more exercises and tests, even though the exercises and tests may be smaller in scope. This course can also be found really cheap ($12-$20) when Udemy runs promotions. While I've done Linux Foundation courses in the past, I'm not a huge advocate of them because I found them less engaging and less practical in regards to real-world experience (not exam experience, but less applicable in the day-to-day work aspect). However, the Linux Foundation CKAD course is the official course by the same organization that provides the certification, so they are always going to be up to date with the latest material that's covered on the exam as the exam changes over time due to new upgrades. It also isn't a bad course or anything like that. It still has quality content. I just found it less engaging than other courses. One thing to note though is that the A Cloud Guru and the Udemy courses are always going to lag behind the Linux Foundation courses because they ACG and Udemy courses have to wait for the exam updates. This CKAD course by the Linux Foundation is also the most expensive of the three courses I recommended, even if it is bundled with the exam. There are occasional deals, but they are less often run. If I had to pick one course to recommend for someone who already has Kubernetes experience, I'd pick the A Cloud Guru one. If I had to pick one for someone with no Kubernetes experience, I'd pick the Udemy one. If you are looking to round out knowledge and just looking for more experience in general and another resource to study, I'd recommend the Linux Foundation course last. Finally, I cannot stress enough how USEFUL the killer.sh mock exams are. The killer.sh mock exams are harder than the cert exam, so the killer.sh mock exams are excellent test preparation. The killer.sh mock exams ask more questions, have more criteria to address per question, and grade you on more components of the question than the cert exams. On top of that, the killer.sh mock exams cover every aspect of the questions that they expect you to be able to answer in depth. They grade you on each of the question's criteria, tell you which parts you got right and wrong, and then provide solutions for you to reference if needed. The killer.sh mock exams are also provided in a very similar manner to what you would find in the real exam. This is a MUST USE resource in my opinion.

@@JarrettCoggin thanks for the recommendation. One question...is the exam content and format changing every 3 months? Will there be any issue if i buy udemy course now, but taking the exam maybe about 4 months

@cffoo438 - The content of the exam keeps pace with Kubernetes as new Kubernetes releases become available, but there's a lag time of 4-8 weeks between when a new version of Kubernetes is released and when the exam is updated after the release. For reference: docs.linuxfoundation.org/tc-docs/certification/faq-cka-ckad-cks#what-application-version-is-running-in-the-exam-environment In regards to the Udemy course I mentioned and as with all Udemy courses, it's up to the course instructor to update the content over time as the material changes. I've found some Udemy courses I've purchased have not been kept up to date as time has gone on. However, the Udemy course I've mentioned claims to be up to date with version 1.26 of the CKAD exam. I went through the content in the second half of 2022 (August is when I picked up the course), so that version of the exam wasn't out yet, but there weren't any discrepancies I found at the time between the course and the exam content at that point in time. On the whole, Kubernetes is not massively changing from release to release, and you can always keep up to date on the changes on the Kubernetes website (kubernetes.io/releases/). Information that I learned in Kubernetes versions 1.15/1.16 is still relevant and useful today, even though there may have been changes in how things are defined in manifest files and the like. I've found the CHANGELOGs on the GitHub repo for Kubernetes to be helpful in understanding what has changed release to release (github.com/kubernetes/kubernetes/tree/master/CHANGELOG). The CHANGELOGs get into some of the finer, more detailed nuances of Kubernetes since they often reference specific pieces that have changed (such as CLI flags changing or being deprecated, specific items moving from beta to stable, etc.). I've also found the Deprecated API Migration Guide helpful in regards to being aware of content that may potentially change in the future (kubernetes.io/docs/reference/using-api/deprecation-guide/). This is something I'd keep an eye on because there was a specific question where I was asked to upgrade an object from a much older version of Kubernetes (say version 1.12) to the current version (1.26). This meant that I had to update the manifest file to bring it current and reapply the object to successfully answer the question. I hope that helps!

@@JarrettCoggin again thanks for the great info. I think i can go ahead and buy the udemy course...and maybe register for exam somewhere around MAY

How's the exam prep going? Have you registered for the exam?

@itlearner1175, thanks for the question! In my experience, you don't need the skills to build your own custom Helm charts for the CKAD exam, but you should have the skills to be able to use existing Helm charts. Custom Controllers was not mentioned in the covered topics for the exam, so I wouldn't worry about them either. You can find the material tested by the exam on this page: training.linuxfoundation.org/certification/certified-kubernetes-application-developer-ckad/

I failed. My struggles 1. Time management: Ran out of time, left 2 QnS. I'm good at CLI. I have been using Linux since decade. 2. how to canary deploy 3. There was one qn ask was to creating cpu limit not beyond the NS QUOTA. but when I can't find any quota on namespace

Thanks for watching! I hope you found it helpful!

what are day to day activities ?