Sysadmin Fundamentals Guide

This is an original planning guide. It is not a vendor manual, an official standard, or legal advice. Example addresses use documentation ranges. Review every command before you paste it.

The boring map of a host

A Unix host has a directory layout you can learn once. /etc is configuration. /var is data that changes, including logs. /home is people. /srv is a common place for service data. /usr is the system, which packages update. /tmp is scratch and not a place to keep a secret or a backup. The directory explorer on this site is that map. The permissions guide says who may write there.

An account is a person or a service. A person has a home and a reason to log in. A service account often should not have an interactive login. Sharing one admin login between people makes the log useless. The career guide talks about how you work. This page only insists the host can tell you apart.

Uptime is how long the OS has been running. An SLA is a promise about a service, often measured in a percentage over a month. They are not the same number. A host can have been up for a year while the service was down. The downtime calculator turns a percentage into minutes so a promise of 99.9 percent becomes a number of minutes you can plan. It does not measure your host.

Change control is a ticket with a before, a change, a check, and a rollback. Every other guide on this desk is that shape applied to one subsystem. If you cannot write the rollback, you are not ready to start, even when the change is small.

A new lab host you can explain in one page

On a new lab host, write the hostname, the OS and version, the purpose, the admin accounts, the service accounts, and where the data lives. Look at /etc, /var, and /srv and confirm the purpose matches a directory you named. A purpose of everything is not a purpose.

Give each admin their own account. The shared password on a wiki is the first thing you remove. Service accounts do not get a login shell if the service does not need one. The permissions on their homes follow the permissions guide, especially .ssh.

Read the uptime. Then write the service promise separately, in minutes per month, using the calculator if you have a percentage. If there is no promise, say so. Do not invent 99.99 percent because it looked professional.

Make one harmless change, such as a message-of-the-day line, with a before and an after. That is the rehearsal for every larger guide. If the rehearsal has no ticket, the larger change will not have one either.

A change you can finish

Fundamentals are the checklist you repeat when a host arrives and when a host changes owners. You do not rebuild a production host to make the directories pretty. You document where it differs and why.

Rollback of the rehearsal change is the previous file. Rollback of a new account is to lock it.

Record hostname, OS version, and purpose.
Name the data directory and confirm it is not /tmp.
Give people separate accounts. Lock shared logins.
Mark service accounts that should not have a shell.
Read uptime, and write the SLA as minutes or as none.
Rehearse one reversible change with a before and after.
Point the next real change at the specialist guide.

Mistakes that outlive the command

sysadmin fundamentals fails in the same few ways. Each one looks finished on the day you type it. The list is here so the ticket can forbid them before the window, not so you can recognise them in an incident review.

One shared root password as the administration method.
Data in /tmp or in a home directory nobody will back up.
Quoting uptime as if it were the service SLA.
A purpose so broad that any process seems allowed.
A change with no rollback because it was small.
Installing a desktop environment on a server to get one graphical tool.
A service account with a login and a reused password.
Skipping the map and opening a specialist guide for a host you cannot name.

Decisions to write down before the window

One decision for sysadmin fundamentals: Hostname, OS version, and purpose fit on one page. Data has a directory. It is not /tmp. People have separate accounts. The password is not shared and not in the ticket. Service accounts have no interactive shell unless the service needs one. Uptime and SLA are both written, and they are not confused. The first change on a new host is reversible and was reversed once in practice or has a written rollback. Specialist work goes to the matching guide after this page is filled in. Backups of the data directory exist or are scheduled before users arrive. The backup guide owns the method. The directory map matches what you saw, not what a different distribution uses. The distro pillars cover those differences. A second person can say what the host is for. And the inventory line matches the hardware guide if this is a physical machine.

Fundamentals checks before the host takes users

Confirm this before you call the change done: Hostname and OS version are recorded. The purpose is one sentence. The data directory exists and is not /tmp. Admin accounts are individual. No shared password is in the ticket. Service accounts that should not log in cannot log in. Uptime was read and not described as the SLA. The SLA is minutes or an explicit none. A rehearsal change has a before and after. Backup is scheduled or the host is still empty. A second person can state the purpose. The next guide is named. The host is the one in the ticket, confirmed by hostname. The hostname in the prompt matches the hostname in the ticket for sysadmin fundamentals. The result line includes the time and the person, not only the command. You can point at the rollback value without searching a chat history. And done is a sentence a second person can repeat: the proof for sysadmin fundamentals happened, and the next host is not started early. If one of those fails, stop.

Tools for this guide

These tools run in the browser on this site. They do not log into your machines.

Guides in this cluster

These supporting pages belong to this pillar.