How do I adjust logging levels in Fusion 5.x? – Lucidworks

Issue:

How do I adjust logging levels in Fusion 5.x?

Environment:

Fusion 5.x

Resolution:

In Fusion 5.x, Fusion Services come pre-configured to write messages to stdout and send messages to Logstash as outlined here.

In a containerized environment like Kubernetes, the recommended approach is to change the values.yaml file used when deploying a Fusion 5.x service. This will set a parameter which sets the logging level for a particular logger.

To update logging to debug for api-gateway, add “logLevelRoot: debug“ to “<cluster>_<namespace>_fusion_values.yaml“ as shown in below code. Make sure it at the same level as “javaToolOptions“.

api-gateway:
 service:
 type: "NodePort"
 ingress:
 enabled: true
 host: "<server>"
 path: "/*"
 tls:
 enabled: true
 javaToolOptions: "-Xmx2g"
 logLevelRoot: debug
 pod:
 annotations:
 prometheus.io/port: "6764"
 prometheus.io/scrape: "true"
 prometheus.io/path: "/actuator/prometheus"

Adjusting individual loggers

By default, loggers are named after the Java package+class name i.e. com.lucidworks.fusion.ml. To adjust the level one or more loggers without changing the overall default, add a parameter to the Java system properties within values.yaml.

ml-model-service:
 enabled: false
 image:
 imagePullPolicy: "IfNotPresent"
 nodeSelector:
 fusion_node_type: system
 modelRepoImpl: fusion
 fs:
 enabled: true
 javaToolOptions: "-Dlogging.level.com.lucidworks.fusion.ml=INFO"

If files are edited and log levels are changed, is a restart of the component required?

In Kubernetes, editing these files is not recommended.

it would have to be done on each pod so if you have three API pods, you’d have to edit each
it requires a restart of the SERVICE process in the pod but not a restart/recreate of the pod.
If the pod is recreated, the edits to the config file are wiped out.

The recommended way to make configuration changes is either to manually edit the Kubernetes objects (Deployments, ReplicaSets, or StatefulSets) or to use environment variables set in the fusion_values.yaml file. Doing either of these methods will result in a restart of all pods managed by that set with the new log settings being applied from that point forward.

Should we enable debugging in Production?

Generally we don’t recommend enabling debugging in production as it can impact performance. This should be ad-hoc on an as-needed basis.