Five mistakes beginners make in DevOps training in Nepal

The most common mistake in DevOps training in Nepal is starting with Kubernetes before you can use Linux. The other four are about cloud costs, keys, backups and the first job title. Each one has a simple fix you can start this week. DevOps means running the servers that apps live on, and sending new code to them safely. Most of the daily work is reading logs and fixing small breakages. Every mistake below makes that work harder later. Why is Kubernetes the wrong place to start DevOps? Kubernetes keeps apps that are packed in containers running on a group of servers. The official Kubernetes tutorial starts from an app that is already in a container. So Kubernetes sits on top of two other skills: Linux and containers. When a pod keeps crashing, you find the cause the same way you would on one Linux server. You read the logs, check the disk and check the port. If tail, grep and df -h are new to you, the Kubernetes error will look like noise. So learn in this order. First Linux and the terminal, then Bash scripts and Git. Then learn how DNS, ports and HTTPS work. Docker comes after that, and Kubernetes after Docker. At Saarathi Academy in Old Baneshwor, Kathmandu, our Cloud Computing and DevOps syllabus follows that order. Weeks 1 to 4 cover Linux, Bash, Git, networking and a safe AWS account. Kubernetes comes in Week 8. Does stopping an AWS server stop the cost? Not fully. AWS's own docs say a stopped EC2 server is not billed for usage, but its EBS storage is still charged. An Elastic IP address also stays yours after you delete the server, until you release it. AWS keeps a list of these surprise charges, and it is worth reading once. The bigger trap is the NAT gateway. Many tutorials add one to a network without a word about cost. AWS charges for it every hour once it exists, even with no traffic. At USD 0.045 an hour in the US East (Ohio) region, a month left idle comes to about USD 32. New AWS accounts start on a Free plan with USD 100 in credits, and you can earn up to USD 100 more. AWS says you are not charged unless you move to a Paid plan. The account closes after 6 months or when the credits run out, whichever comes first. So on the Free plan, a forgotten resource does not charge your dollar card. It uses up your credits, and your practice account can close weeks early. Stay on the Free plan, and do not upgrade just to make one lab work. The fix is a habit. Set a budget alert on the first day. When a lab ends, delete the server, do not just stop it. Then check the Volumes page and your Elastic IPs before you log out. In our course, Week 4 sets budget alarms at USD 1, 5 and 20, and every Terraform lab in Week 7 is graded on terraform destroy. What happens if you push an AWS key to GitHub? Beginners often paste an access key into a script to make a lab work, then push it to GitHub. GitHub now turns on push protection by default for personal accounts. It stops a push to a public repository when it spots a secret it recognises. That helps, but it only catches the patterns it knows. The better habit is to never create keys for your AWS root account. Give each tool its own role with only the access it needs. Keep every key out of files that Git tracks. In our syllabus, Week 6 builds pipelines that log in to AWS with a short-lived role through OIDC, so no key is stored at all. The same week uses gitleaks to scan the code for secrets. Is a backup you never restored really a backup? No. Many beginners set up a nightly database dump with cron and stop there. Later they find the file was empty, or nobody wrote down how to restore it. A backup counts only after you have restored it once on a fresh machine and checked the data. This is slow, boring work. It is also the work a company needs on the day a disk fails. In Week 2 of our course, you install PostgreSQL, run pg_dump on a schedule with rotation, and then test a restore. Why do beginners aim for the wrong first DevOps job? Many people finish a short course and apply only to "DevOps Engineer" posts. Read those posts closely. A DevOps Engineer post from Nepal Clearing House on merojob asked for 3 to 5 years of experience, with at least 2 years as a DevOps engineer. That is not a reason to give up. It means the first job has a different title. Look for internships, junior cloud support, system administrator and IT support roles where you work on Linux servers every day. Then grow into DevOps over one to two years. We say this plainly in our own syllabus. The goal of the course is an internship or a junior cloud support job, with no salary promise. Most DevOps jobs ask for 2 or more years of experience. One project on GitHub helps more than a long list of tool names on your CV. Put one small app online yourself, with HTTPS, a pipeline and a README that a stranger could follow. What can you try tonight? You do not need a server or a cloud account to start. On Windows 10 (version 2004 or later) or Windows 11, open PowerShell as administrator and run wsl --install. Microsoft's WSL install guide says this sets up Ubuntu. Restart, open Ubuntu, then run df -h, free -h and ls -l /var/log. Find out what each one tells you about your machine. For reading, the book The Linux Command Line is a free PDF on linuxcommand.org. Work through its first part before you pay for any DevOps training. If the terminal starts to feel normal, you are ready for the next step. The week-by-week Cloud Computing and DevOps syllabus and the course page are public before anyone pays.