MTU Problems — Diagnosing Fragmentation and Packet Size Issues¶
MTU (Maximum Transmission Unit) defines the largest packet size that can be transmitted over a network interface without fragmentation. Incorrect MTU settings can cause packet fragmentation, application timeouts, VPN failures, slow network performance, Kubernetes connectivity issues, TLS handshake failures, and "black-hole" connections. Understanding MTU is essential for troubleshooting modern enterprise, cloud, container, and hybrid networking environments. Every Network Engineer, Linux Administrator, DevOps Engineer, SRE, Cloud Architect, and Kubernetes Administrator should understand MTU troubleshooting.
Learning Path¶
Course Progress
What You'll Learn¶
After completing this lesson, you'll be able to:
- Understand MTU
- Learn IP fragmentation
- Diagnose MTU mismatches
- Understand Path MTU Discovery (PMTUD)
- Troubleshoot VPN and cloud MTU issues
- Analyze fragmentation problems
- Optimize MTU in production environments
Prerequisites¶
Complete:
Basic understanding of:
- Ethernet
- IP Packets
- Transmission Control Protocol (TCP)
Why Do MTU Problems Occur?¶
Imagine users report:
or
or
Possible causes:
- MTU Mismatch
- Packet Fragmentation
- PMTUD Failure
- Firewall Blocking ICMP
- VPN Overhead
What is MTU?¶
MTU stands for:
It defines:
Ethernet MTU¶
Standard Ethernet MTU:
This is the default value on most networks.
Jumbo Frames¶
Some data centres use:
Benefits:
- Lower CPU Usage
- Higher Throughput
- Better Storage Performance
Common for:
- Storage Area Network (SAN)
- Network-Attached Storage (NAS)
- High-Speed Networks
Packet Flow¶
The packet must fit within the MTU of every link along the path.
Packet Fits¶
Example:
MTU:
Result:
Packet Too Large¶
Example:
MTU:
Result:
or
depending on configuration.
IP Fragmentation¶
Large packets may be divided into:
The receiving host reassembles the fragments.
Problems with Fragmentation¶
Fragmentation causes:
- Increased CPU Usage
- More Packets
- Reduced Performance
- Higher Packet Loss Risk
- Slower Applications
Modern networks generally try to avoid fragmentation.
Don't Fragment (DF) Bit¶
IPv4 packets may include:
If set:
If the packet exceeds the MTU:
An ICMP message should be returned indicating that fragmentation is needed.
Path MTU Discovery (PMTUD)¶
PMTUD automatically determines the smallest MTU along the network path.
Workflow:
Applications then transmit packets using the discovered MTU.
PMTUD Failure¶
Sometimes firewalls block ICMP.
Result:
This is called an:
MTU Black Hole¶
Symptoms:
- HTTPS Fails
- SSH Freezes
- VPN Disconnects
- Large File Transfers Fail
- Small Packets Work
These issues are often difficult to diagnose.
VPN MTU Problems¶
VPN encapsulation adds extra headers.
Example:
↓
VPN Header
↓
If the tunnel MTU is smaller:
The tunnel MTU must account for encapsulation overhead.
Kubernetes MTU¶
Container networking introduces additional encapsulation.
Examples:
- VXLAN
- Geneve
- Generic Routing Encapsulation (GRE)
Effective MTU becomes smaller.
Typical values:
depending on the Container Network Interface (CNI) plugin and encapsulation method.
Cloud MTU¶
Cloud providers typically use:
Some services support:
Always verify:
- VPC/VNet MTU
- VPN MTU
- Overlay Network MTU
Check MTU¶
Linux:
Example:
Change MTU¶
Temporary change:
Verify:
Test MTU with Ping¶
Linux:
1472 bytes + 28-byte IP/ICMP headers = 1500 bytes.
If the packet succeeds:
If it fails:
Reduce the payload size until it succeeds.
Windows MTU Test¶
-fsets the Don't Fragment flag.-lspecifies the payload size.
Detect Fragmentation¶
Capture packets.
Look for:
- Fragmented IP Packets
- ICMP "Fragmentation Needed"
- Retransmissions
Wireshark Analysis¶
Look for:
- Fragmented Packets
- ICMP Type 3 Code 4
- Retransmissions
- TCP Timeouts
Useful filters:
or
Kubernetes Perspective¶
Verify:
- Pod MTU
- CNI MTU
- Overlay Network MTU
- VXLAN Configuration
Example:
inside a Pod or node.
Enterprise Troubleshooting Workflow¶
Common MTU Values¶
| Network | Typical MTU |
|---|---|
| Ethernet | 1500 |
| Jumbo Frame | 9000 |
| VXLAN Overlay | 1450 |
| GRE Tunnel | ~1476 |
| IPsec VPN | ~1400–1438 |
| WireGuard VPN | ~1420 |
Exact values vary depending on encapsulation overhead and implementation.
CLI Examples¶
View MTU.
Change MTU.
Test MTU.
Capture fragmented packets.
Hands-on Lab¶
Task 1¶
View the MTU of all interfaces.
Task 2¶
Determine the largest packet that can be sent without fragmentation.
Reduce the payload size until the test succeeds.
Task 3¶
Temporarily change the MTU.
Test connectivity again.
Task 4¶
Capture packets during an MTU test.
Observe fragmentation behavior.
Task 5¶
Open the packet capture in Wireshark.
Identify:
- Fragmented Packets
- ICMP Fragmentation Needed Messages
- Retransmissions
Task 6¶
Deploy a VPN tunnel in a lab and identify the optimal MTU.
Task 7¶
Inspect the MTU configuration used by your Kubernetes CNI plugin.
Task 8¶
Draw the packet flow:
Explain what happens when the packet exceeds the MTU of one router along the path.
Production Troubleshooting¶
Problem:
Check:
- MTU
- PMTUD
- ICMP
- VPN
- Fragmentation
- Overlay Network
- Firewall
- Packet Capture
Workflow:
Fragmentation vs PMTUD¶
| Fragmentation | PMTUD |
|---|---|
| Splits Large Packets | Finds the Best MTU |
| Higher Overhead | Optimized Packet Size |
| Less Efficient | Better Performance |
| Legacy Approach | Preferred Modern Approach |
| Can Increase Packet Loss | Reduces Fragmentation |
MTU Problems vs Routing Issues¶
| MTU Problem | Routing Issue |
|---|---|
| Large Packets Fail | All Traffic May Fail |
| Fragmentation | Missing Route |
| PMTUD Failure | Routing Loop |
| Black-Hole Connections | Destination Unreachable |
| Packet Size Issue | Path Selection Issue |
Common Mistakes¶
❌ Assuming MTU is always 1500.
✅ Verify MTU on every network segment.
❌ Blocking ICMP.
✅ Allow PMTUD-related ICMP messages.
❌ Ignoring VPN overhead.
✅ Reduce tunnel MTU appropriately.
❌ Using Jumbo Frames on unsupported devices.
✅ Ensure end-to-end Jumbo Frame support.
❌ Overlooking CNI MTU settings.
✅ Verify overlay network MTU in Kubernetes.
Best Practices¶
- Keep MTU consistent across connected networks where possible.
- Allow ICMP messages required for PMTUD.
- Avoid unnecessary fragmentation.
- Validate MTU after VPN deployment.
- Test MTU after cloud network changes.
- Configure overlay networks with appropriate MTU values.
- Monitor retransmissions and fragmentation.
- Document MTU settings for production environments.
Interview Questions¶
Beginner¶
- What is MTU?
- What is the default Ethernet MTU?
- What is fragmentation?
- What is the DF bit?
Intermediate¶
- Explain Path MTU Discovery.
- What is an MTU black hole?
- How do VPNs affect MTU?
- How do you identify MTU problems?
Architect Level¶
- Design an MTU troubleshooting workflow for a hybrid cloud environment.
- Explain why Kubernetes overlay networks require smaller MTU values.
- How would you troubleshoot intermittent HTTPS failures caused by MTU mismatches?
Summary¶
In this lesson, you learned:
- MTU
- Maximum Transmission Unit
- IP Fragmentation
- Path MTU Discovery (PMTUD)
- MTU Black Holes
- VPN MTU
- Kubernetes MTU
- Packet Fragmentation Analysis
- Production MTU Troubleshooting
MTU configuration plays a critical role in network reliability and performance. Incorrect MTU values can cause subtle and difficult-to-diagnose issues such as intermittent application failures, VPN instability, and fragmented traffic. By understanding PMTUD, fragmentation, and packet capture analysis, engineers can efficiently diagnose and resolve MTU-related problems across enterprise, cloud, and Kubernetes environments.
Key Takeaways¶
- MTU defines the largest packet that can traverse a network without fragmentation.
- The standard Ethernet MTU is 1500 bytes.
- PMTUD helps determine the optimal packet size for a path.
- Blocking ICMP can cause MTU black-hole issues.
- VPNs and overlay networks reduce the effective MTU because of encapsulation.
- Use Ping, tcpdump, and Wireshark together to diagnose MTU-related problems.
What's Next?¶
In the next lesson, you'll learn about Latency.
You'll explore:
- Network Latency
- Round Trip Time (RTT)
- Jitter
- Throughput
- Bandwidth vs Latency
- Latency Measurement
- Production Performance Troubleshooting
By the end of the lesson, you'll understand how to measure, analyse, and reduce latency across enterprise networks, cloud platforms, and Kubernetes environments.