diff --git a/docs/en/observability/profiling-known-issues.asciidoc b/docs/en/observability/profiling-known-issues.asciidoc index eed109dc59..aa517cb4a9 100644 --- a/docs/en/observability/profiling-known-issues.asciidoc +++ b/docs/en/observability/profiling-known-issues.asciidoc @@ -3,21 +3,20 @@ Universal Profiling has the following known issues: -//// -TEMPLATE -Note: Add known issues for newer Elastic Stack -versions to the top of this page - [discrete] -== Brief description +== Helm chart `profiling-collector` faulty binary location + +_Affected: 8.16.0, 8.16.1, 8.16.2, 8.17.0_ +_Fixed in: 8.16.3+, 8.17.1+_ + +The installation of Universal Profiling collector using the Helm chart `profiling-collector` produces a Deployment with +the wrong binary path set in `spec.template.spec.containers['pf-elastic-collector'].command`. -_Versions: XX.XX.XX, YY.YY.YY, ZZ.ZZ.ZZ_ +The correct path should be `/home/nonroot/pf-elastic-collector` while the `command` field references the deprecated path +`/root/pf-elastic-collector`. This issue prevents the collector from starting and as a result, the Deployment has Pods +in `CrashLoopBackoff` status. -// Detailed description including: +To fix this issue, a manual edit of the Deployment manifest should be performed, removing entirely the `command` field +from the `spec` section. -// The conditions in which this issue occurs -// The behavior of the issue -// Why it happens -// If applicable, exact error messages linked to this issue so users searching for the error message end up here -// If applicable, link to fix -//// \ No newline at end of file +This can be performed by either manipulating the Deployment manifest directly or by using the `kubectl edit` command.