Skip to main content

Pod network corruption

Last updated on

Pod network corruption is a Kubernetes pod-level chaos fault that flips random bits in a configurable percentage of packets on the network path serving a target pod for a configurable duration. Corrupted packets fail their IP, TCP, or UDP checksums and are silently discarded by the receiver, so corruption manifests to the application as packet loss but with normal round-trip time. Only the selected pods are affected; other pods on the node and the node's host networking are unaffected.

Use this fault to test how a service behaves when the network mangles bytes on the wire: a faulty NIC, a degraded fiber link, MTU mismatches with fragmentation issues, or a misbehaving overlay encapsulation.

Run your first experiment

If you have not configured the chaos infrastructure yet, go to Quickstart to install the chaos infrastructure and run an experiment end to end.

Use cases​

Run this fault when you want to answer concrete questions like:

  • Checksum-based retransmit: Does TCP recover from corrupted packets via retransmission, or does the connection stall?
  • Application-level integrity: For protocols that carry their own checksums (databases, RPCs, file transfers), does the application detect corruption and refuse the data, or does it silently accept bad bytes?
  • TLS handshake under corruption: TLS records are checksum-protected; what happens to the handshake when 25% of packets get corrupted? Does the client retry cleanly?
  • gRPC and HTTP/2 framing: Are gRPC and HTTP/2 framed correctly enough to drop only the corrupted frames, or does corruption tear down the whole connection?

Prerequisites​

  • Kubernetes version: 1.21 or later. Go to What's supported to confirm distribution support.
  • Target pods are Running: The application pods you intend to target are in the Running state before the fault is launched.
  • Privileged pods allowed: The cluster lets you schedule privileged pods in the chaos namespace.
  • Workload selector defined: The chaos experiment knows the target workload by kind, namespace, and either names or labels.

Supported environments​

PlatformSupport status
Amazon EKSSupported
Azure AKSSupported
Google GKESupported
Red Hat OpenShiftSupported
RancherSupported
VMware TanzuSupported
Self-managed Kubernetes (CNCF-certified)Supported
GKE AutopilotSupported with Autopilot setup
EKS Fargate, ACI virtual nodesNot supported (no access to container runtime sockets)

Permissions required​

The fault runs under the chaos infrastructure's service account.

Resource (apiGroup)VerbsWhy it is needed
pods ("")get, list, create, delete, deletecollection, patch, updateDiscover target pods and run the chaos pod on the same node
pods/log ("")get, list, watchStream chaos pod logs for status and debugging
deployments, statefulsets, replicasets, daemonsets (apps)get, listResolve the target workload to the pods it owns
events ("")get, list, create, patch, updateRecord fault progress as Kubernetes events
jobs (batch)get, list, create, delete, deletecollectionRun the chaos job that drives the fault

The default Harness chaos infrastructure service account already includes these permissions.

Fault tunables​

Configure the following fault parameters when you add Pod network corruption to an experiment in Chaos Studio. Defaults are shown for reference.

Chaos parameters

TunableDescriptionDefault
NETWORK_PACKET_CORRUPTION_PERCENTAGEPercentage of packets to corrupt on the matched flows. 100 corrupts every packet; 25 flips bits on a quarter of them.100
CORRELATIONCorruption correlation as a percentage. 0 makes each packet independent; higher values clump corrupt packets.""
TOTAL_CHAOS_DURATIONDuration of the fault in seconds.60

Traffic filters

TunableDescriptionDefault
DESTINATION_IPSComma-separated list of destination IPs. Corruption applies only to packets headed to these IPs. Empty matches all destinations.""
DESTINATION_HOSTSComma-separated list of destination hostnames. The helper resolves them and adds the resolved IPs to the filter.""
SOURCE_PORTSComma-separated list of source ports on the target pod. Empty matches all source ports.""
DESTINATION_PORTSComma-separated list of destination ports. Empty matches all destination ports.""
NETWORK_INTERFACENetwork interface inside the target container's namespace. Almost always eth0 for standard CNI plugins.eth0

Targeting

TunableDescriptionDefault
TARGET_PODSComma-separated list of pod names to target. Empty selects from the workload's pods using POD_AFFECTED_PERCENTAGE.""
TARGET_CONTAINERContainer in the pod whose network namespace to enter. Empty targets the first container in the pod spec.""
NODE_LABELLabel selector to filter target pods by the node they run on. Empty disables node-based filtering.""
POD_AFFECTED_PERCENTAGEPercentage of the workload's pods to target. 0 means one pod.0
SEQUENCEWhen multiple pods are targeted, inject parallel (all at once) or serial (one after another).parallel

Runtime and helper

TunableDescriptionDefault
CONTAINER_RUNTIMEContainer runtime on the target nodes. One of containerd, docker, crio.containerd
SOCKET_PATHPath to the container runtime socket on the target node. Set to match CONTAINER_RUNTIME./run/containerd/containerd.sock
RAMP_TIMEWait period in seconds before and after the fault. Go to ramp time to read how it is applied.0

Tunables that apply to every chaos fault are documented in common tunables for all faults.

Configure for your container runtime​

Set CONTAINER_RUNTIME and SOCKET_PATH to match the runtime on the target node:

CONTAINER_RUNTIMESOCKET_PATH
containerd (default)/run/containerd/containerd.sock
docker/var/run/docker.sock
crio/var/run/crio/crio.sock

Fault execution in brief​

Configures the container's network interface to corrupt a specified percentage of packets (flipping a random bit in each), optionally scoping the effect to only certain destination IPs, hosts, or ports so other traffic passes through unaffected.

Expected behavior during fault execution​

  • TCP connections retransmit corrupted segments. At low corruption (10%), throughput drops modestly. At high corruption (50+%), the connection effectively halts because every retransmit is also likely corrupted.
  • UDP traffic (DNS, QUIC, video) loses corrupted packets without retransmission. Application-level logic determines whether the call fails or recovers.
  • TLS handshakes can fail if the corruption rate exceeds the renegotiation tolerance of the TLS stack. Modern stacks typically retry the handshake a few times before giving up.
  • gRPC and HTTP/2 detect framing errors and may tear down the entire connection; the client typically reconnects.
  • Health probes that route through the corrupted path can fail; the kubelet may mark the pod NotReady.
When the fault ends

The corruption configuration is removed, and the pod's network returns to normal immediately. Active TCP connections resume retransmits without corruption and recover.

Signals to watch​

Attach resilience probes to assert each layer:

  • TCP retransmit and error counters: Use a Prometheus probe on node_netstat_Tcp_RetransSegs or your CNI's equivalent.
  • Application error rate: Use an HTTP probe for endpoint health; checksum failures appear as connection errors, not slow responses.
  • Pod readiness: Use a Kubernetes probe to fail when the target pod stays NotReady longer than your SLO.

Verify the fault execution effect​

While the experiment is running, confirm corruption is reaching the application:

  1. Look for TCP retransmits inside the target container.

    kubectl exec -n <namespace> <pod-name> -c <target-container> -- cat /proc/net/snmp | grep -A 1 ^Tcp:

    RetransSegs should rise sharply during the fault as receivers discard corrupted segments and senders retransmit.

  2. Drive traffic from another pod and watch for errors.

    kubectl run -n <namespace> tester --image=nicolaka/netshoot --rm -it -- \
      curl -m 5 http://<target-pod-ip>:<port>/healthz

    At high NETWORK_PACKET_CORRUPTION_PERCENTAGE, requests fail or stall on retransmits. At lower values, latency rises while throughput drops.

Recovery and cleanup​

  • End of duration: The corruption configuration is removed automatically and connectivity returns to normal within a few seconds.
  • Abort the experiment: Stopping the experiment from Chaos Studio triggers the same cleanup path.
  • Failed cleanup: If automated cleanup did not complete, restart the target pod to reset its network state.

Limitations​

  • Serverless Kubernetes (EKS Fargate, ACI virtual nodes): These platforms do not allow the privileged access this fault needs. GKE Autopilot is supported once the one-time setup in Chaos on GKE Autopilot is in place.
  • Windows containers: This fault is supported on Linux pods only.
  • CNI plugins that bypass the pod's eth0: Custom CNI plugins routing host-side may not be affected by this fault.
  • hostNetwork pods: The fault refuses to inject on hostNetwork: true pods to avoid corrupting host traffic.

Troubleshooting​

Pod network corruption experiment stays Pending or never starts in Harness Chaos Engineering

Inspect the chaos pods in the experiment namespace with kubectl describe pod -n <chaos-namespace>. The most common causes are taints, insufficient resources, or PodSecurity admission blocking privileged pods. Add the required tolerations to the experiment or run in a namespace with privileged Pod Security level.

No application errors observed during pod-network-corruption

The most common causes are: corruption percentage too low to overwhelm TCP retransmits (try 50% or higher); the filter is too narrow and matches no real traffic (broaden DESTINATION_IPS/HOSTS/PORTS); or NETWORK_INTERFACE does not match the pod's interface. Drive traffic from another pod with curl or ping and watch for retransmits on the target container's /proc/net/snmp.

Connection to container runtime fails for pod-network-corruption in Harness Chaos Engineering

The default SOCKET_PATH is /run/containerd/containerd.sock. For Docker, set CONTAINER_RUNTIME=docker and SOCKET_PATH=/var/run/docker.sock. For CRI-O, set CONTAINER_RUNTIME=crio and SOCKET_PATH=/var/run/crio/crio.sock.

Corruption persists after pod-network-corruption ends

Automated cleanup did not complete. Restart the target pod to reset its network state. If the issue recurs, capture the chaos pod logs from the experiment namespace before the next run and share them with Harness support.