This Splunk OpenTelemetry Collector for Kubernetes release adopts the Splunk OpenTelemetry Collector v0.147.0.
🛑 Breaking changes 🛑
-
operator: Change defaultoperator.admissionWebhooks.autoGenerateCert.recreatefromtruetofalse. (#2261)
The self-signed webhook certificate is now reused across helm upgrades instead of being
regenerated every time. The default certificate validity is 10 years (certPeriodDays: 3650),
so rotation per-upgrade is not required for most deployments.
To restore the previous behavior (regenerate certificate on every upgrade), set:operator: manager: rolling: true # recommended admissionWebhooks: autoGenerateCert: recreate: true
Note:
manager.rolling: trueis recommended withrecreate: true— it restarts the operator
pod so it loads the new certificate promptly. Without it, there is a delay before the
webhook server picks up the new cert, during which Instrumentation CR requests may fail
with x509 errors. -
agent: Dropdestination_service_namespacefrom Istio metric attributes (#2275)
Users running the chart withautodetectenabled have hit the dimension limit
for Istio metrics. This fix drops thedestination_service_namespacemetric
dimension to avoid hitting the dimension limit and dropping the data point entirely.
💡 Enhancements 💡
operator: Add Job-based Instrumentation CR installation Job. (#2261)
The Instrumentation CR is deployed as a regular Helm resource by default. A new option
instrumentation.installationJob.enableddeploys the CR via a Kubernetes Job that waits
for webhook readiness before applying the CR, with built-in retry. This is the recommended
option for Helm v4 or environments with slow operator startup.
See Auto-instrumentation Install for details.
🧰 Bug fixes 🧰
eks_addon: Fixed journald receiver issues in the EKS Add-on by mounting required shared library paths used by journalctl and aligning the directory structure with EKS node conventions. (#2260)