Today, console log entries and FFDC log entries are getting routed to the Kubernetes logging server (like LogDNA or CloudWatch), but not the trace log entries. I have it so you can specify the trace spec in my CR yaml, but right now, you can only see that if you kubectl exec into the pod and do a cat /logs/trace.log. Need to adjust the WLP_LOGGING_CONSOLE_SOURCE to include trace entries, in each microservice's yaml in my helm chart (that gets wrapped by my operator).
To be more explicit, this:
{{- if .Values.global.jsonLogging }}
- name: WLP_LOGGING_CONSOLE_FORMAT
value: json
- name: WLP_LOGGING_CONSOLE_SOURCE
value: message,ffdc
{{- end }}
Needs to become this:
{{- if .Values.global.jsonLogging }}
- name: WLP_LOGGING_CONSOLE_FORMAT
value: json
- name: WLP_LOGGING_CONSOLE_SOURCE
value: message,ffdc,trace
{{- end }}
I could consider adding http access logs too - need to decide if that would be useful. Note all of this special routing of logs other than the console log is only available in Liberty when JSON logging mode is enabled.
Today, console log entries and FFDC log entries are getting routed to the Kubernetes logging server (like LogDNA or CloudWatch), but not the trace log entries. I have it so you can specify the trace spec in my CR yaml, but right now, you can only see that if you
kubectl exec
into the pod and do acat /logs/trace.log
. Need to adjust theWLP_LOGGING_CONSOLE_SOURCE
to include trace entries, in each microservice's yaml in my helm chart (that gets wrapped by my operator).To be more explicit, this:
Needs to become this:
I could consider adding http access logs too - need to decide if that would be useful. Note all of this special routing of logs other than the console log is only available in Liberty when JSON logging mode is enabled.