Skip to main content
OTel Fleet Monitor for AWS app icon

OTel Fleet Monitor for AWS

Inventory, health and configuration drift for OpenTelemetry Collectors on Amazon EKS and ECS, directly in Splunk Platform. Finds silent collectors, dropped telemetry, outdated versions, plaintext secrets and differences between expected and applied collector configuration.Built by zemour Ali
splunk product badge

Default Version 1.2.0

November 14, 2026

Compatibility

Splunk Enterprise, Splunk Cloud

Platform Version: 10.8, 10.7, 10.6, 10.5, 10.4, 10.3, 10.2, 10.1, 10.0, 9.4, 9.3, 9.2, 9.1, 9.0

Rating
0
(0)

Log in to rate this app

Support
Not Supported

OpenTelemetry Collectors are now the backbone of telemetry pipelines on AWS, but most teams cannot answer basic questions about their own fleet: How many collectors run on EKS and ECS? Which ones stopped sending data? Which ones drop spans, metrics or logs? Which versions are outdated? Is the configuration that actually runs the one we expect? OTel Fleet Monitor for AWS answers these questions in Splunk Platform (Splunk Enterprise and Splunk Cloud Platform), without requiring Splunk Observability Cloud. 1. Inventory A modular input calls read-only AWS APIs (ECS, EKS, SSM, STS) and the read-only Kubernetes API of each EKS cluster to discover every collector: ECS services and tasks, DaemonSets, Deployments, StatefulSets, pods, ConfigMaps, OpenTelemetryCollector custom resources and EKS add-ons such as ADOT. Clusters without any collector are reported. 2. Health The inventory is cross-checked with the collector internal telemetry (otelcol_* metrics): - Silent collectors: running on AWS but no longer reporting (SILENT, NEVER_REPORTED, NOT_INVENTORIED). - Dropped data: refused, dropped and failed-to-export items by stage, signal and component, plus exporter sending-queue saturation. - Versions: CURRENT, OUTDATED, OBSOLETE, UNSUPPORTED or UNPINNED (latest tag), based on a version policy fed by GitHub releases and on AWS add-on versions. 3. Expected vs applied configuration Define expected rules (version, config hash, required or forbidden components) per account, cluster, namespace or workload. The app compares them with the configuration really applied, detects instances still running an older configuration (previous task definition revision, pods started before the last ConfigMap change) and flags bad practices such as a missing memory_limiter, disabled retry or sending queue, disabled TLS verification and plaintext secrets. Included: 8 dashboards, 9 alerts (shipped disabled), KV store reference data, outbound proxy support. All AWS and Kubernetes calls are read-only; secrets are stored encrypted in Splunk and never indexed.