Cluster logging 5.6.0 breaks log forwarding with loki
theSuess opened this issue · 2 comments
theSuess commented
Describe the bug
After upgrading to 5.6.0, cluster log forwarding no longer works when using the default
output in combination with loki
Environment
- OCP 4.11, cluster-logging 5.6.0
apiVersion: logging.openshift.io/v1
kind: ClusterLogForwarder
metadata:
name: instance
namespace: openshift-logging
spec:
inputs: []
outputs:
- name: fluentbit
type: fluentdForward
url: 'tcp://fluentbit.custom-logging.svc.cluster.local:24224'
pipelines:
- inputRefs:
- audit
name: audit-logs
outputRefs:
- fluentbit
parse: json
- inputRefs:
- application
name: application-logs
outputRefs:
- fluentbit
- default
Expected behavior
The logs are forwarded to loki and fluentbit alike
Actual behavior
The configuration fails because of unrecognized outputs: ["default"]
To Reproduce
Steps to reproduce the behavior:
- Configure Loki as log storage
- use the provided ClusterLogForwarder
- Observe the failure
Additional context
I assume the issue stems from #1731. Reading the code, the way to go would be specifying an empty output of type Loki, but this is only supported for the Elasticsearch output.
Our current workaround is to specify the default output manually
jcantrill commented
closing in favor of https://issues.redhat.com/browse/LOG-3437