-
Bug
-
Resolution: Done-Errata
-
Normal
-
Logging 5.8.3
-
False
-
None
-
False
-
NEW
-
VERIFIED
-
-
Bug Fix
-
-
-
Log Storage - Sprint 250, Log Storage - Sprint 251
-
Moderate
Description of problem:
The status of a LokiStack contains a set of Conditions that signal the overall status of the managed Loki deployment. One of those conditions is the ReadyCondition which should signal that all components are up and ready to receive requests.
Currently this condition is based only on the running-state of the pods, which causes the ReadyCondition to be emitted by the operator as soon as all components are running.
Because this is not immediately obvious to the user, it is possible for them to miss a configuration issue which causes one of the components to not get ready or a component not getting ready after a restart.
How reproducible:
Deploy a LokiStack and observe both the readiness states of the created Pods and the conditions in the LokiStack.
Actual results:
The ReadyCondition has Status=True as soon as all Pods are showing "Running" as status, even when some containers still are unready. Usually the ingester Pod is the last one to get ready.
Expected results:
The ReadyCondition should only have Status=True when all the Pods report all containers ready.
- clones
-
LOG-5160 Loki Operator should not emit ReadyCondition when readiness checks fail
- Closed
- links to
-
RHSA-2024:129168 Logging for Red Hat OpenShift - 5.8.5
- mentioned on