Anyone who's worked with CentOS networking knows the frustration: you carefully configure your /etc/resolv.conf with optimized parameters like:

options rotate timeout:1 attempts:2
nameserver 8.8.8.8
nameserver 8.8.4.4
nameserver 1.1.1.1

Only to find it mysteriously reset after a reboot or network service restart. This happens because CentOS's network management stack (NetworkManager or legacy network scripts) regenerates this file.

The default DNS resolution behavior in Linux has several shortcomings:

• Serial querying (no rotation) of nameservers
• 5-second timeout per query
• Only 2 retries before failing

This creates a worst-case scenario where DNS resolution can take up to 10 seconds (2 retries × 5s timeout) before failing over to the next nameserver.

For modern CentOS systems using NetworkManager:

1. Create or edit /etc/NetworkManager/conf.d/dns.conf:

[main]
dns=none

2. Set up a static resolv.conf:

# chattr +i /etc/resolv.conf

3. Alternative method - configure DNS options in NetworkManager:

nmcli connection modify eth0 ipv4.dns-options "rotate timeout:1 attempts:2"
nmcli connection modify eth0 ipv4.dns "8.8.8.8 8.8.4.4 1.1.1.1"
nmcli connection up eth0

For systems using traditional network scripts:

# Add to /etc/sysconfig/network-scripts/ifcfg-eth0:
DNS1=8.8.8.8
DNS2=8.8.4.4
DNS3=1.1.1.1
PEERDNS=no
RES_OPTIONS="rotate timeout:1 attempts:2"

On newer CentOS versions with systemd-resolved:

# /etc/systemd/resolved.conf
[Resolve]
DNS=8.8.8.8 8.8.4.4 1.1.1.1
DNSOverTLS=opportunistic
Cache=yes
DNSStubListener=yes

Then enable and start the service:

systemctl enable systemd-resolved
systemctl start systemd-resolved
ln -sf /run/systemd/resolve/stub-resolv.conf /etc/resolv.conf

Test your DNS resolution behavior with:

dig example.com | grep "Query time"

Or for a more thorough test:

for i in {1..10}; do 
  time dig example.com @8.8.8.8 >/dev/null
done

---

Anyone who's worked with CentOS networking knows the frustration: you carefully configure /etc/resolv.conf with optimized DNS settings like rotate, timeout, and attempts, only to find your changes mysteriously wiped out after a reboot or network service restart.

# Typical resolv.conf optimizations that get lost
options timeout:2 attempts:3 rotate
nameserver 8.8.8.8
nameserver 8.8.4.4
nameserver 1.1.1.1

This behavior stems from CentOS's network manager services (NetworkManager or dhclient) which regenerate resolv.conf dynamically. The default settings are particularly problematic because:

• DNS queries use sequential rather than round-robin rotation
• 5-second timeout per query is too long for failover scenarios
• Only 2 retries before giving up

For CentOS/RHEL 7+ (Using NetworkManager)

Create or modify /etc/NetworkManager/conf.d/dns.conf:

[main]
dns=none

[connection]
dns=8.8.8.8,8.8.4.4,1.1.1.1
dns-options=rotate,timeout:2,attempts:3

Then apply changes:

systemctl restart NetworkManager

For Systems Using dhclient

Add to /etc/dhcp/dhclient.conf:

supersede domain-name-servers 8.8.8.8, 8.8.4.4;
supersede domain-search "example.com";
prepend domain-name-servers 1.1.1.1;
timeout 2;
retry 3;

After making changes, test with:

# Check current DNS servers
systemd-resolve --status

# Test query rotation
dig example.com | grep SERVER
# Run multiple times to see different servers responding

For complete control, you can make /etc/resolv.conf immutable:

chattr +i /etc/resolv.conf

Remember this may break DHCP client operations that expect to modify the file.

With these optimizations, DNS failover time improves dramatically:

Setting	Default	Optimized
Timeout	5s	2s
Attempts	2	3
Worst-case failover	10s	6s

The rotate option provides load balancing across all configured nameservers, not just failover.