Loading...

XML

Word

Printable

Details

Type: Bug
Resolution: Done
Priority: Critical
Fix Version/s: Logging 5.7.1
Affects Version/s: Logging 5.7.0
Component/s: Log Collection
Labels:
- devel_ack+

Blocked:
False
Blocked Reason:
None
Ready:
False
Docs QE Status:
NEW
QE Status:
NEW
Release Note Text:

Hide
Before this change the API server would default the new collector type field to "vector" even when the custom resource still wanted to use the old field. This change removes the default for the new collector type field in order to restore the previous behavior

Show
Before this change the API server would default the new collector type field to "vector" even when the custom resource still wanted to use the old field. This change removes the default for the new collector type field in order to restore the previous behavior
Release Note Type:
Bug Fix

Sprint:
Log Collection - Sprint 236

SFDC Cases Links:
SFDC Cases Counter:

Description

On a cluster partially managed by app-sre via app-interface, the RHOBS team had deployed the cluster logging operator and set up this configuration for the cluster logging resource: https://gitlab.cee.redhat.com/service/app-interface/-/blob/master/resources/setup/cluster-logging/clusterlogging.yaml.

The cluster log forwarder was configured with the highlighted lines from the following file: https://gitlab.cee.redhat.com/service/app-interface/-/blob/master/resources/rhobs/logs/clusterlogforwarder.yaml.j2#L8-26

Upon the automatic upgrade of the cluster-logging operator from 5.6 to 5.7, it was noticed that all the pods in the logging collector daemonset were chrashlooping.

Here are some details about the crashlooping pods:

They were running Vector, surprisingly, even though we had configured fluentd as collector.
The following errors appeared in the events:
- pipeline_0_: