"Temporary failure in name resolution": fix DNS on a Linux server

1. Is it DNS, or the whole network?

ping -c 3 1.1.1.1
ip route

• No reply from 1.1.1.1 either → it’s not DNS. The server has no route out: check ip route for a default via … line, the interface (ip -br a), the cloud firewall / security group.
• The IP replies, names don’t → it’s DNS. Continue.

(Some networks block ping. If so, try curl -sI https://1.1.1.1 instead.)

2. Which DNS server does the system use?

cat /etc/resolv.conf

On Ubuntu and many Debian systems you’ll see:

nameserver 127.0.0.53
options edns0 trust-ad

127.0.0.53 is systemd-resolved, a local DNS service that forwards questions to the real servers. The real ones are here:

resolvectl status

Link 2 (eth0)
    Current Scopes: DNS
Current DNS Server: 10.0.0.2
       DNS Servers: 10.0.0.2

No DNS Servers line at all means the network configuration gave this interface no DNS servers. If systemd-resolved itself isn’t running, 127.0.0.53 answers nothing:

systemctl status systemd-resolved --no-pager
sudo systemctl restart systemd-resolved

On RHEL-family systems, /etc/resolv.conf is usually written by NetworkManager and lists the real servers directly; nmcli dev show | grep DNS shows them too.

3. Ask a DNS server directly

dig github.com                   # through the system's configured servers
dig @10.0.0.2 github.com         # one specific server
dig @1.1.1.1 github.com          # a public one, for comparison

(dig is in bind9-dnsutils on Debian/Ubuntu, bind-utils on RHEL. nslookup github.com 1.1.1.1 works too.)

4. Set DNS servers that stick

Don’t just edit /etc/resolv.conf: on most systems it’s generated, and your change disappears at the next reboot or DHCP renewal. Set the servers where the network is configured.

Ubuntu (netplan): in the file under /etc/netplan/, under your interface:

network:
  version: 2
  ethernets:
    eth0:
      dhcp4: true
      nameservers:
        addresses: [1.1.1.1, 9.9.9.9]

sudo netplan try        # applies it, and rolls back by itself if you lose the connection

RHEL / Rocky / AlmaLinux (NetworkManager):

nmcli connection show                                  # find the connection name
sudo nmcli connection modify "System eth0" ipv4.dns "1.1.1.1 9.9.9.9"
sudo nmcli connection up "System eth0"

On a cloud server, prefer the provider’s own DNS address if it has one (for example 169.254.169.253 on AWS, or the .2 address of the VPC) — internal names often resolve only there.

5. Inside Docker containers

Containers get their own copy of the resolver settings. If the host uses 127.0.0.53 (an address that means nothing inside a container), Docker gives the container the host’s real upstream servers instead, or public ones as a last resort. On company networks that block public DNS, containers then fail while the host works. Tell Docker which servers to use in /etc/docker/daemon.json:

{ "dns": ["10.0.0.2", "1.1.1.1"] }

sudo systemctl restart docker

Recreate the containers afterwards (docker compose up -d --force-recreate). Check from inside one: docker run --rm busybox nslookup github.com.