Skip to main content

Pod JVM Mongo latency

Last updated on

Pod JVM Mongo latency is a Kubernetes pod-level chaos fault that adds a configurable delay to MongoDB driver calls from a JVM running in a target container, scoped to a chosen database, collection, and operation, for a configurable duration. Only matched calls are slowed; unrelated MongoDB operations and other code paths run at normal speed. When the fault ends, MongoDB calls return to baseline latency immediately.

Use this fault to test how a Java service behaves when MongoDB becomes slow on a specific code path: a primary under heavy load, a long-running aggregation, or a cross-region replica taking longer to ack.

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:

  • Driver timeout sensitivity: When a MongoDB find takes 2 seconds instead of 20 ms, does the application surface the slow query through tracing, or does it block worker threads?
  • Connection-pool saturation: Does the application's connection pool back-pressure correctly or starve under slow queries?
  • Aggregation tail latency: Slow only aggregate operations and verify whether the application's reporting paths cope with the delay.
  • Write-path back-pressure: Slow insert or update and see whether the application throttles producers or queues writes in memory.
  • Caller circuit breaker: Does a circuit breaker around the slow operation trip and recover correctly when the fault ends?

Prerequisites​

  • Kubernetes version: 1.21 or later. Go to What's supported to confirm distribution support.
  • Target pod is Running: The Java application pod is in the Running state.
  • Java agent attach available: The Java process allows agent attach. Utilities such as ps, pgrep, and bash are present in the container, and the JVM is not built with a restricted runtime that strips attach modules.
  • MongoDB Java driver in classpath: The target JVM uses the MongoDB Java driver and exercises the configured DATABASE, COLLECTION, and METHOD.
  • Privileged pods allowed: The cluster lets you schedule privileged pods in the chaos namespace. GKE Autopilot supports this fault but requires the one-time setup in Chaos on GKE Autopilot; other locked-down distributions may need similar exemptions.
  • Container runtime access: The chaos pod can reach the container runtime socket on the target node (/run/containerd/containerd.sock, /var/run/docker.sock, or /var/run/crio/crio.sock).
  • Workload selector defined: The chaos experiment knows the target workload by kind, namespace, and either names or labels.
JVM chaos uses the Byteman agent

This fault attaches a Byteman agent to the target JVM over BYTEMAN_PORT. The port must be reachable from the chaos pod and must not be in use by the application.

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 JVM Mongo latency to an experiment in Chaos Studio. Defaults are shown for reference.

MongoDB filters

TunableDescriptionDefault
DATABASETarget MongoDB database name. Empty matches all databases.""
COLLECTIONTarget MongoDB collection name. Empty matches all collections in the database.""
METHODMongoDB operation to target. Common values: find, insert, update, delete, aggregate."find"
TRANSACTION_PERCENTAGEPercentage of matched MongoDB operations to delay, between 0 and 100. 0 delays none; 100 delays every match.0

Chaos parameters

TunableDescriptionDefault
LATENCYDelay to add to each matched operation, in milliseconds.2000
TOTAL_CHAOS_DURATIONDuration of the fault in seconds.60

JVM

TunableDescriptionDefault
BYTEMAN_PORTPort on which the Byteman agent listens inside the container. Must not conflict with any port already in use.9091
JAVA_HOMEAbsolute path to the Java installation inside the container. Empty auto-detects from PATH.""

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 running the JVM. 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

Common pod selection tunables (TARGET_WORKLOAD_KIND, TARGET_WORKLOAD_NAMESPACE, TARGET_WORKLOAD_NAMES, TARGET_WORKLOAD_LABELS) are documented in common pod fault tunables. Tunables that apply to every fault are documented in common tunables for all faults.

Match operation to scenario

For read-heavy workloads, target find and aggregate. For ingest paths, target insert and update. Scoping the slow operation matches a production failure mode and keeps the blast radius small.

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​

Attaches a Java agent to the target JVM and intercepts MongoDB driver operations matching DATABASE, COLLECTION, and METHOD to add LATENCY milliseconds to each matched call on the configured percentage, for TOTAL_CHAOS_DURATION seconds.

Expected behavior during fault execution​

  • Matched MongoDB operations take longer by approximately LATENCY ms. Other operations and unrelated collections run normally.
  • Caller-side metrics (request latency, queue depth, connection-pool waiters) rise to reflect the added delay.
  • Clients with timeouts shorter than LATENCY cancel the call and may retry.
  • Thread-pool-bound applications saturate quickly if many concurrent callers wait on the slow operation.
  • Tracing systems show the matched driver span growing by LATENCY ms.
When the fault ends

MongoDB calls return to baseline latency immediately. Calls in flight finish at the delayed time and then the system returns to normal.

Signals to watch​

Attach resilience probes to assert each layer:

  • Driver-level latency: Use a Prometheus probe on mongodb.driver.commands duration or your APM's MongoDB latency metric.
  • Caller timeouts: Use an HTTP probe against endpoints that read or write to the targeted collection.
  • Pod readiness: Use a Kubernetes probe to fail when the target pod oscillates NotReady.

Verify the fault execution effect​

While the experiment is running, confirm operations are slower:

  1. Time a request that exercises the matched operation.

    kubectl run -n <namespace> tester --image=nicolaka/netshoot --rm -it -- \
      curl -w "time=%{time_total}\n" -o /dev/null -s http://<service>:<port>/<endpoint>

    time_total should rise by approximately LATENCY ms when the endpoint exercises the matched operation.

  2. Confirm in tracing.

    The driver span for the matched operation should be approximately LATENCY ms longer than its baseline.

Recovery and cleanup​

  • End of duration: MongoDB calls return to baseline latency automatically.
  • Abort the experiment: Stopping the experiment from Chaos Studio triggers the same cleanup path.
  • Stuck threads: If the application is wedged because of saturated thread or connection pools, restart the pod.

Limitations​

  • Serverless Kubernetes (EKS Fargate, ACI virtual nodes): These platforms do not expose container runtime sockets and reject the privileged access the 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.
  • Non-JVM and non-Mongo workloads: This fault targets the MongoDB Java driver inside a JVM.
  • Reactive drivers: Reactive MongoDB driver variants may surface latency differently from blocking drivers depending on the application's subscription handling.

Troubleshooting​

Pod JVM Mongo latency 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 on the target node that the chaos pods do not tolerate, insufficient resources, or a PodSecurity admission policy blocking privileged pods. Add the required tolerations or run in a namespace with privileged Pod Security level.

No latency observed during pod-jvm-mongo-latency

The most common causes are: DATABASE or COLLECTION does not match; METHOD does not match the driver operation; TRANSACTION_PERCENTAGE is 0 (default) so nothing is intercepted; or the application uses a reactive driver path. Re-run with TRANSACTION_PERCENTAGE=100, empty COLLECTION, and a larger LATENCY to confirm the path is working.

Connection to container runtime fails for pod-jvm-mongo-latency 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.