# How to set the time zone, hostname and NTP on a Linux server

> Set a server's time zone with timedatectl, give it a proper hostname, and keep its clock in sync with chrony or systemd-timesyncd, then check it really is.

Source: https://vpsnine.com/tutorials/set-timezone-and-ntp-linux · Updated: 2026-10-21

Difficulty: beginner · Time: about 10 minutes · Tested on: Ubuntu 24.04 LTS, Debian 13 Stable, AlmaLinux 10

Prerequisites:

- A Linux server with systemd
- Root or a user with sudo

To set the time zone on a Linux server, run `sudo timedatectl set-timezone Europe/Amsterdam` (any name from `timedatectl list-timezones`), or `UTC`. To check the clock is kept in sync, run `timedatectl` and look for `System clock synchronized: yes` and `NTP service: active`; if it isn't, install `chrony` and enable it.

The hostname is set with `sudo hostnamectl set-hostname web-1`. All three settings take effect at once, with no reboot. This tutorial was run on Ubuntu 24.04, Debian 13 and AlmaLinux 10; see the note under each step for what a container could and couldn't test.

## Check the current settings

```bash
timedatectl
```

```text
               Local time: Wed 2026-10-21 14:32:10 UTC
           Universal time: Wed 2026-10-21 14:32:10 UTC
                 RTC time: Wed 2026-10-21 14:32:10
                Time zone: Etc/UTC (UTC, +0000)
System clock synchronized: yes
              NTP service: active
          RTC in local TZ: no
```

The two lines that matter are **Time zone** and **System clock synchronized**.

## Set the time zone

Find your zone's name, then set it:

```bash
timedatectl list-timezones | grep -i amsterdam
sudo timedatectl set-timezone Europe/Amsterdam
```

Zone names are `Region/City`, such as `America/New_York` or `Asia/Singapore`. A misspelled one is refused with `Failed to set time zone: Invalid or not installed time zone`.

### UTC or local time?

Many people leave servers on UTC: logs from servers in different places line up, and there's no daylight saving jump that runs a 02:30 job twice or not at all. Pick local time if the people reading logs or the cron schedules think in it. Either way, applications should store timestamps in UTC; the server's zone only changes how times are shown and when cron fires.

To go back to UTC:

```bash
sudo timedatectl set-timezone UTC
```

### Check it worked

`date` now prints the time with your zone's abbreviation (`CEST`, `EDT`), and `/etc/localtime` links to the zone file. Restart long-running services that print times (`sudo systemctl restart cron`, your app) so they pick up the new zone; new processes see it immediately.

## Set the hostname

```bash
sudo hostnamectl set-hostname web-1
```

Use a short name of letters, digits and hyphens. It's stored in `/etc/hostname` and shows up in your shell prompt (after you log in again), in logs and in emails the server sends.

Some programs look the hostname up and wait for a timeout if they can't resolve it (`sudo` warns `unable to resolve host web-1`). Add it to `/etc/hosts`:

```text
127.0.1.1   web-1.example.com web-1
```

On Ubuntu and Debian `127.0.1.1` is the usual line for this; on AlmaLinux and Rocky, append the names to the existing `127.0.0.1` line instead.

The hostname is a label for the server itself. What the internet sees is DNS: point a name at the server with an [A record](https://vpsnine.com/tutorials/point-a-domain-to-your-vps), and for a mail server, set [reverse DNS](https://vpsnine.com/help/reverse-dns) to match.

On a cloud server, cloud-init may set the hostname and rewrite `/etc/hosts` again at every boot, undoing both changes. Tell it to leave them alone:

```bash
sudo tee /etc/cloud/cloud.cfg.d/99-hostname.cfg <<'EOF'
preserve_hostname: true
manage_etc_hosts: false
EOF
```

We couldn't test `hostnamectl set-hostname` in a container (containers can't change their own hostname); the commands are the documented ones from `man hostnamectl`.

## Keep the clock in sync with NTP

A server's clock drifts by seconds a week without correction. TLS certificates, two-factor codes, database replication and log correlation all need the right time. NTP (Network Time Protocol) fixes it continuously.

Most images already run an NTP client:

| Distribution | Default client |
| --- | --- |
| Ubuntu 24.04 | `systemd-timesyncd` (simple, client only) |
| Debian 12 and 13 | Depends on the image; `timedatectl` and `systemctl status chrony systemd-timesyncd` tell you |
| AlmaLinux, Rocky Linux, Fedora | `chronyd` (package `chrony`) |

Either is fine for a server. `chrony` handles bad network conditions better and can serve time to other machines; `timesyncd` is smaller. If `timedatectl` says `NTP service: active` and `synchronized: yes`, you're done.

### Install chrony

```bash
sudo apt install -y chrony                                          # Ubuntu, Debian
sudo dnf install -y chrony && sudo systemctl enable --now chronyd   # AlmaLinux, Rocky
```

On Ubuntu and Debian, installing `chrony` removes `systemd-timesyncd` (the packages conflict, since only one should set the clock) and starts `chrony` right away.

The servers it asks are in `/etc/chrony/chrony.conf` (Ubuntu, Debian) or `/etc/chrony.conf` (AlmaLinux, Rocky): `pool` lines pointing at the distribution's NTP pool. The defaults are fine. To use a provider's time servers, add a `server` or `pool` line and restart the service (`chrony` on Ubuntu and Debian, `chronyd` on AlmaLinux and Rocky).

### Check it worked

```bash
chronyc tracking
chronyc sources -v
```

In `tracking`, **System time** is how far the clock is from NTP time, normally well under a millisecond once it has settled, and **Leap status** should be `Normal`. In `sources`, the line starting with `^*` is the server currently in use.

To see the offset against one server without changing anything, chrony has a query-only mode (the service needn't be running):

```bash
sudo chronyd -Q 'server time.cloudflare.com iburst'
```

It prints `System clock wrong by -0.541810 seconds (ignored)` or similar, then exits. That's what we ran in our containers, which share the host's clock and so can't set their own.

## Troubleshooting

**`System clock synchronized: no` for more than a few minutes.** The NTP client can't reach its servers. NTP uses outbound UDP port 123; check that a firewall isn't blocking it, then `chronyc sources` (all `^?` means unreachable) or `journalctl -u systemd-timesyncd`.

**`NTP service: n/a`.** No NTP client that `timedatectl` knows about is installed. Install `chrony` as above.

**`sudo: unable to resolve host web-1: Name or service not known`.** The new hostname isn't in `/etc/hosts`. Add it as shown above.

**Cron jobs ran at the wrong hour after changing the zone.** Cron read the zone when it started. `sudo systemctl restart cron` (`crond` on AlmaLinux and Rocky).

## Next steps

- Schedule jobs in the zone you just set: [cron vs systemd timers](https://vpsnine.com/tutorials/cron-vs-systemd-timers).
- Read logs with times in that zone: [journalctl](https://vpsnine.com/tutorials/read-logs-with-journalctl).
- The rest of a new server's setup: the [first-hour checklist](https://vpsnine.com/help/first-hour-checklist).

## Doing it on VPSNine

On VPSNine, the hostname you type when you order a server is set by cloud-init on first boot, with `/etc/hosts` managed to match; that's why the `99-hostname.cfg` step above matters if you rename it later. The panel keeps showing the name you ordered with. Images start on UTC. Every [plan](https://vpsnine.com/plans) runs [Ubuntu 24.04](https://vpsnine.com/os/ubuntu-24.04), [Debian 13](https://vpsnine.com/os/debian-13) and [AlmaLinux 10](https://vpsnine.com/os/almalinux-10).