-
Feature
-
Resolution: Unresolved
-
Normal
-
None
-
None
-
False
-
None
-
False
-
Not Selected
-
0
Proposed title of this feature request
OpenShift logging to support to renaming labels via vector
What is the nature and description of the request?
Currently log filtering with openshift logging does only support send logs to different outputs based on namespace or pod labels.
There seems to be no support for more advanced filters.
what customer wants is to change some metadata labels of openshift log message, because of their internal requirements.
current cluster logforwarder configuration:
spec:
inputs:
- application:
namespaces:
- demo
name: demo-logs
outputs:
- loki:
labelKeys:
- openshift.labels.job
- log_type
- streamer
- kubernetes.namespace_name
- kubernetes.pod_name
- kubernetes.container_name
- kubernetes.labels.app
- kubernetes.labels.cs-dice/application
- kubernetes.labels.cs-dice/lane
- kubernetes.labels.job-name
- kubernetes.labels.release
tenantKey: kubernetes.namespace_name
name: loki-logs
type: loki
url: 'http://...:3100'
pipelines:
- inputRefs:
- demo-logs
labels:
job: openshift-logging-vector
name: loki-logs
outputRefs:
- loki-logs
for example the label "kubernetes.labels.cs-dice/application" is currently converted to "kubernetes_labels_cs-dice_application" and need to rewrite this as just "application".
As per customer, currently this is not supported with openshift logging 5.8.
Another issue is that currently there is no way of separating logs based on whether they were written to stdout or stderr.
It would be nice if there is a field in the json representation of the log message clearly stating if the message was received on stdout or stderr.
Why does the customer need this? (List the business requirements)
when moving to vector label names change which breaks their current logging pipeline
List any affected packages or components.
Openshift logging