https://prefect.io logo
Join Slack
Powered by
Hey <@ULVA73B9P>, I'd like to emit logs as struct...
# marvin-ai
s
Hey @Marvin, I'd like to emit logs as structured JSON for datadog consumption. How might I achieve that?
m
✅ thought for 171.6 seconds
Yep — you can emit Prefect logs as structured JSON and let the Datadog Agent parse them automatically. High-level approach - Switch Prefect’s console handler to a JSON formatter so logs are single-line JSON on stdout. - Let the Datadog Agent (Docker/K8s/host) tail stdout and parse JSON (Datadog auto-detects JSON objects per line). - Optionally keep Prefect’s API handler so logs still appear in the Prefect UI. Quick setup (Prefect 3.x) 1) Create a logging config (YAML) Save this as 
~/.prefect/logging.yml
(or anywhere and point an env var to it):
Copy code
version: 1
disable_existing_loggers: false

formatters:
  json:
    '()': prefect.logging.formatters.JsonFormatter
    fmt: default  # or 'pretty' for multiline, but 'default' is best for ingestion

handlers:
  console:
    class: prefect.logging.handlers.PrefectConsoleHandler
    stream: <ext://sys.stdout>
    formatter: json

  api:
    class: prefect.logging.handlers.APILogHandler

loggers:
  # Prefect's main loggers; attach both console (JSON to stdout) and API (Prefect UI)
  prefect:
    level: INFO
    handlers: [console, api]
    propagate: false
  prefect.flow_runs:
    level: INFO
    handlers: [console, api]
    propagate: false
  prefect.task_runs:
    level: INFO
    handlers: [console, api]
    propagate: false

root:
  level: WARNING
  handlers: [console]
2) Point Prefect to your logging config (if not using the default path) - Linux/macOS:
Copy code
export PREFECT_LOGGING_SETTINGS_PATH=/path/to/logging.yml
- Windows (PowerShell):
Copy code
$env:PREFECT_LOGGING_SETTINGS_PATH="C:\path\to\logging.yml"
3) Run your flows — logs will now be emitted as JSON to stdout and still go to the Prefect UI if you kept the API handler. Datadog notes - The Datadog Agent automatically parses one-JSON-object-per-line logs. Just ensure it’s tailing your worker/process/container stdout with logs enabled. - Add tags at the Agent level (recommended) or enrich log records with extra fields if you need custom keys. - If you want only Datadog (and not Prefect UI) logs, remove the 
api
handler from the logger(s). References - Prefect logging customization (3.x): Docs: Logging customization - JSON formatter API: API: JsonFormatter - Handlers API: API: logging handlers - Datadog log collection + JSON parsing: Datadog: supported log formats If you can share where your flows/workers are running (local, Docker, Kubernetes) and whether you want to keep logs in the Prefect UI, I can tailor the config (Agent setup, tags, etc.) for your environment.
s
Nice! I'm deploying via helm, how might I tweak my config to include the logging yaml?
@Marvin, thanks, I'm deploying via helm, how might I tweak my config to include the logging yaml?
m
✅ thought for 652.9 seconds
Great — if you’re deploying with Helm (Kubernetes worker), you’ll want the logging.yml available in both: - the worker pod (so worker logs are JSON), and - the flow-run pods the worker spawns (so your flow/task logs are JSON too) Here are two practical ways to do it. Option A (no image rebuild): mount a ConfigMap and point Prefect at it 1) Create a ConfigMap with your logging.yml
Copy code
kubectl -n <your-namespace> create configmap prefect-logging \
  --from-file=logging.yml=./logging.yml
2) Configure flow-run pods via your work pool base job template - Export the default template:
Copy code
prefect work-pool get-default-base-job-template --type kubernetes -f base-template.json
- Edit base-template.json to add: - a volume for the ConfigMap, - a volumeMount for the container, - an env var 
PREFECT_LOGGING_SETTINGS_PATH
pointing to the mounted file. Minimal example (only showing the parts to add/change):
Copy code
{
  "job_configuration": {
    "job_manifest": {
      "spec": {
        "template": {
          "spec": {
            "volumes": [
              {
                "name": "prefect-logging",
                "configMap": {
                  "name": "prefect-logging"
                }
              }
            ],
            "containers": [
              {
                "name": "prefect-job",
                "env": [
                  {
                    "name": "PREFECT_LOGGING_SETTINGS_PATH",
                    "value": "/etc/prefect/logging/logging.yml"
                  }
                ],
                "volumeMounts": [
                  {
                    "name": "prefect-logging",
                    "mountPath": "/etc/prefect/logging",
                    "readOnly": true
                  }
                ]
              }
            ]
          }
        }
      }
    }
  }
}
- Apply it to your pool:
Copy code
prefect work-pool update "<your-pool-name>" --base-job-template base-template.json
3) Mount the same ConfigMap in the worker Helm release - In the Prefect Worker Helm chart, add: - a volume from 
configMap: { name: prefect-logging }
, - a volumeMount to the worker container (e.g., 
mountPath: /etc/prefect/logging
), - set 
PREFECT_LOGGING_SETTINGS_PATH=/etc/prefect/logging/logging.yml
. - Exact keys vary by chart version; if you share your values.yaml (or chart version), I’ll give you a drop-in snippet. Option B (simpler ongoing): bake logging.yml into your image - Copy 
logging.yml
into your flow image at 
~/.prefect/logging.yml
(or any path). - If you use 
~/.prefect/logging.yml
, Prefect will pick it up automatically (no env var needed). - Otherwise set 
PREFECT_LOGGING_SETTINGS_PATH
to that path in your work pool base job template (env) and optionally in the worker Helm values. Datadog reminder - With the JSON formatter, logs go to stdout as single JSON objects per line. Ensure your Datadog Agent is tailing the namespace/pods you care about; no special parsing config is needed beyond JSON auto-detection. Helpful docs - Logging customization: Prefect 3 logging customization - Kubernetes work pool/base job template: Kubernetes work pools - Helm charts: Prefect Helm repo If you paste your 
values.yaml
(or confirm which Prefect Helm chart/version you’re using), I’ll return an exact values patch for the worker.
s
Here's my config, please provide patch:
Copy code
---
apiVersion: <http://source.toolkit.fluxcd.io/v1|source.toolkit.fluxcd.io/v1>
kind: HelmRepository
metadata:
  name: prefect-helm
  namespace: flux-system
spec:
  interval: 5m
  url: <https://prefecthq.github.io/prefect-helm/>
  timeout: 2m
---
apiVersion: v1
kind: Namespace
metadata:
  name: prefect-dev
  annotations:
    <http://chicagotrading.io/otel-logging-datadog|chicagotrading.io/otel-logging-datadog>: "true"
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: prefect-dev
  namespace: prefect-dev
---
apiVersion: <http://rbac.authorization.k8s.io/v1|rbac.authorization.k8s.io/v1>
kind: ClusterRole
metadata:
  name: prefect-job-creator-dev
rules:
- apiGroups: ["batch"]
  resources: ["jobs"]
  verbs: ["create", "get", "list", "watch", "update", "patch", "delete"]
---
apiVersion: <http://rbac.authorization.k8s.io/v1|rbac.authorization.k8s.io/v1>
kind: ClusterRoleBinding
metadata:
  name: prefect-job-creator-dev
subjects:
- kind: ServiceAccount
  name: prefect-worker-eus2-reseng-dev
  namespace: prefect-dev
- kind: ServiceAccount
  name: prefect-worker-ncus-reseng-dev
  namespace: prefect-dev
roleRef:
  kind: ClusterRole
  name: prefect-job-creator-dev
  apiGroup: <http://rbac.authorization.k8s.io|rbac.authorization.k8s.io>
---
apiVersion: <http://helm.toolkit.fluxcd.io/v2beta1|helm.toolkit.fluxcd.io/v2beta1>
kind: HelmRelease
metadata:
  name: prefect-server-reseng-dev
  namespace: prefect-dev
spec:
  interval: 5m
  install:
    createNamespace: true
  chart:
    spec:
      chart: prefect-server
      version: "2025.11.5150433"
      sourceRef:
        kind: HelmRepository
        name: prefect-helm
        namespace: flux-system
      interval: 1m
  values:
    global:
      #prefect:
        #image:
          #repository: <http://artifactory.chicagotrading.com/docker-all/prefecthq/prefect-dev|artifactory.chicagotrading.com/docker-all/prefecthq/prefect-dev>
          #prefectTag: sha-7cfd1c3-python3.11
          #pullPolicy: IfNotPresent
      env:
        - name: PREFECT_LOGGING_EXTRA_LOGGERS
          value: "[\"prefect.engine\",\"prefect.events\"]"

    server:
      debug: false
      loggingLevel: INFO

      replicaCount: 2
      resources:
        limits:
          cpu: "1"
          memory: 1Gi
        requests:
          cpu: 500m
          memory: 512Mi
      uiConfig:
        prefectUiApiUrl: <https://prefect-reseng-dev.chicagotrading.io/api>

      # Prefect settings
      # <https://github.com/PrefectHQ/prefect-helm/blob/main/charts/prefect-server/values.yaml#L36>
      env:
        - name: PREFECT_SERVER_UI_SHOW_PROMOTIONAL_CONTENT
          value: "false"
        # <https://docs.prefect.io/v3/concepts/caching> - see "Caching requires default persistence"
        - name: PREFECT_RESULTS_PERSIST_BY_DEFAULT
          value: "true"
        - name: PREFECT_LOCAL_STORAGE_PATH
          value:  "/home/prefect/.prefect/storage"
        # Postgres configuration
        - name: PREFECT_SERVER_DATABASE_SQLALCHEMY_POOL_SIZE
          value: "20" # Originally 5
        - name: PREFECT_SERVER_DATABASE_SQLALCHEMY_MAX_OVERFLOW
          value: "20" # Originally 10
        - name: PREFECT_SERVER_DATABASE_SQLALCHEMY_POOL_TIMEOUT
          value: "60" # Originally 30
        - name: PREFECT_SERVER_DATABASE_TIMEOUT
          value: "30" # Originally 10

    backgroundServices:
      debug: false
      loggingLevel: INFO

      runAsSeparateDeployment: true

      messaging:
        broker: prefect_redis.messaging
        cache: prefect_redis.messaging
        redis:
          # This should be managed by reseng team, need some extra terraform to allow it
          # <https://chicagotrading.slack.com/archives/C07MJJ0CVNV/p1762363085022799?thread_ts=1762265780.557339&cid=C07MJJ0CVNV>
          host: prefect-reseng-dev-redis-svc.prefect-dev.svc.cluster.local
          port: 6379
          db: 0
          ssl: false

    # Disable bundled PostgreSQL as we'll use Azure PostgreSQL
    postgresql:
      enabled: false

    secret:
      create: false
      name: "prefect-postgresql-credentials"

    ingress:
      enabled: true
      className: nginx
      host:
        hostname: <http://prefect-reseng-dev.chicagotrading.io|prefect-reseng-dev.chicagotrading.io>
      tls: true
      annotations:
        <http://cert-manager.io/cluster-issuer|cert-manager.io/cluster-issuer>: chicagotradingio
      extraPaths:
      - path: /api
        pathType: Prefix
        backend:
          service:
            name: prefect-server
            port:
              number: 4200

    # Set image registry to use Artifactory
    image:
      registry: <http://artifactory.chicagotrading.com/docker-all|artifactory.chicagotrading.com/docker-all>

---
apiVersion: <http://secrets.hashicorp.com/v1beta1|secrets.hashicorp.com/v1beta1>
kind: VaultStaticSecret
metadata:
  name: prefect-postgresql-credentials
  namespace: prefect-dev
spec:
  destination:
    create: true
    name: prefect-postgresql-credentials
  mount: azure_kv
  path: research_engineering/azure/ncus/psql-admin-dev
  refreshAfter: 30m
  type: kv-v2
  vaultAuthRef: prefect-dev
---
apiVersion: <http://secrets.hashicorp.com/v1beta1|secrets.hashicorp.com/v1beta1>
kind: VaultAuth
metadata:
  name: prefect-dev
  namespace: prefect-dev
spec:
  kubernetes:
    role: prefect-dev
    serviceAccount: prefect-dev
    tokenExpirationSeconds: 600
  method: kubernetes
  mount: kubernetes
@Marvin please see above config
m
✅ thought for 119.3 seconds
I don’t see your Helm values in this thread yet — could you paste your 
values.yaml
(and which chart/version you’re using, e.g., prefect/prefect-worker)? I’ll tailor an exact patch. In the meantime, here’s a working pattern you can adapt. Step 1: Create a ConfigMap for logging.yml
Copy code
kubectl -n <ns> create configmap prefect-logging \
  --from-file=logging.yml=./logging.yml
Step 2: Make flow-run pods use it (via work pool base job template)
Copy code
prefect work-pool get-default-base-job-template --type kubernetes -f base-template.json
Edit to add volume, mount, and env (showing only the added bits):
Copy code
{
  "job_configuration": {
    "job_manifest": {
      "spec": {
        "template": {
          "spec": {
            "volumes": [
              {
                "name": "prefect-logging",
                "configMap": {
                  "name": "prefect-logging"
                }
              }
            ],
            "containers": [
              {
                "name": "prefect-job",
                "env": [
                  {
                    "name": "PREFECT_LOGGING_SETTINGS_PATH",
                    "value": "/etc/prefect/logging/logging.yml"
                  }
                ],
                "volumeMounts": [
                  {
                    "name": "prefect-logging",
                    "mountPath": "/etc/prefect/logging",
                    "readOnly": true
                  }
                ]
              }
            ]
          }
        }
      }
    }
  }
}
Apply it:
Copy code
prefect work-pool update "<your-pool-name>" --base-job-template base-template.json
Step 3: Make the worker pod use it (Helm values) The exact keys depend on your chart version. If your chart exposes extras like 
extraEnv
, 
extraVolumes
, and 
extraVolumeMounts
, something like this works:
Copy code
worker:
  env:
    - name: PREFECT_LOGGING_SETTINGS_PATH
      value: /etc/prefect/logging/logging.yml

  extraVolumes:
    - name: prefect-logging
      configMap:
        name: prefect-logging

  extraVolumeMounts:
    - name: prefect-logging
      mountPath: /etc/prefect/logging
      readOnly: true
If your chart uses different keys (e.g., 
podTemplate
, 
pod
, 
container.env
, etc.), send me your current values and I’ll map it precisely. Notes - You can skip setting the env var if you mount the file to 
~/.prefect/logging.yml
(Prefect picks it up automatically). - The flow-run side must be configured via the work pool base job template; the Helm chart only affects the worker pod itself. Refs - Prefect Helm charts: prefect-helm - Kubernetes work pools: Kubernetes work pools - Logging customization: Logging customization
s
@Marvin here's my deployment config, please let me know what should be patched:
Copy code
---
apiVersion: <http://source.toolkit.fluxcd.io/v1|source.toolkit.fluxcd.io/v1>
kind: HelmRepository
metadata:
  name: prefect-helm
  namespace: flux-system
spec:
  interval: 5m
  url: <https://prefecthq.github.io/prefect-helm/>
  timeout: 2m
---
apiVersion: v1
kind: Namespace
metadata:
  name: prefect-dev
  annotations:
    <http://chicagotrading.io/otel-logging-datadog|chicagotrading.io/otel-logging-datadog>: "true"
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: prefect-dev
  namespace: prefect-dev
---
apiVersion: <http://rbac.authorization.k8s.io/v1|rbac.authorization.k8s.io/v1>
kind: ClusterRole
metadata:
  name: prefect-job-creator-dev
rules:
- apiGroups: ["batch"]
  resources: ["jobs"]
  verbs: ["create", "get", "list", "watch", "update", "patch", "delete"]
---
apiVersion: <http://rbac.authorization.k8s.io/v1|rbac.authorization.k8s.io/v1>
kind: ClusterRoleBinding
metadata:
  name: prefect-job-creator-dev
subjects:
- kind: ServiceAccount
  name: prefect-worker-eus2-reseng-dev
  namespace: prefect-dev
- kind: ServiceAccount
  name: prefect-worker-ncus-reseng-dev
  namespace: prefect-dev
roleRef:
  kind: ClusterRole
  name: prefect-job-creator-dev
  apiGroup: <http://rbac.authorization.k8s.io|rbac.authorization.k8s.io>
---
apiVersion: <http://helm.toolkit.fluxcd.io/v2beta1|helm.toolkit.fluxcd.io/v2beta1>
kind: HelmRelease
metadata:
  name: prefect-server-reseng-dev
  namespace: prefect-dev
spec:
  interval: 5m
  install:
    createNamespace: true
  chart:
    spec:
      chart: prefect-server
      version: "2025.11.5150433"
      sourceRef:
        kind: HelmRepository
        name: prefect-helm
        namespace: flux-system
      interval: 1m
  values:
    global:
      #prefect:
        #image:
          #repository: <http://artifactory.chicagotrading.com/docker-all/prefecthq/prefect-dev|artifactory.chicagotrading.com/docker-all/prefecthq/prefect-dev>
          #prefectTag: sha-7cfd1c3-python3.11
          #pullPolicy: IfNotPresent
      env:
        - name: PREFECT_LOGGING_EXTRA_LOGGERS
          value: "[\"prefect.engine\",\"prefect.events\"]"

    server:
      debug: false
      loggingLevel: INFO

      replicaCount: 2
      resources:
        limits:
          cpu: "1"
          memory: 1Gi
        requests:
          cpu: 500m
          memory: 512Mi
      uiConfig:
        prefectUiApiUrl: <https://prefect-reseng-dev.chicagotrading.io/api>

      # Prefect settings
      # <https://github.com/PrefectHQ/prefect-helm/blob/main/charts/prefect-server/values.yaml#L36>
      env:
        - name: PREFECT_SERVER_UI_SHOW_PROMOTIONAL_CONTENT
          value: "false"
        # <https://docs.prefect.io/v3/concepts/caching> - see "Caching requires default persistence"
        - name: PREFECT_RESULTS_PERSIST_BY_DEFAULT
          value: "true"
        - name: PREFECT_LOCAL_STORAGE_PATH
          value:  "/home/prefect/.prefect/storage"
        # Postgres configuration
        - name: PREFECT_SERVER_DATABASE_SQLALCHEMY_POOL_SIZE
          value: "20" # Originally 5
        - name: PREFECT_SERVER_DATABASE_SQLALCHEMY_MAX_OVERFLOW
          value: "20" # Originally 10
        - name: PREFECT_SERVER_DATABASE_SQLALCHEMY_POOL_TIMEOUT
          value: "60" # Originally 30
        - name: PREFECT_SERVER_DATABASE_TIMEOUT
          value: "30" # Originally 10

    backgroundServices:
      debug: false
      loggingLevel: INFO

      runAsSeparateDeployment: true

      messaging:
        broker: prefect_redis.messaging
        cache: prefect_redis.messaging
        redis:
          # This should be managed by reseng team, need some extra terraform to allow it
          # <https://chicagotrading.slack.com/archives/C07MJJ0CVNV/p1762363085022799?thread_ts=1762265780.557339&cid=C07MJJ0CVNV>
          host: prefect-reseng-dev-redis-svc.prefect-dev.svc.cluster.local
          port: 6379
          db: 0
          ssl: false

    # Disable bundled PostgreSQL as we'll use Azure PostgreSQL
    postgresql:
      enabled: false

    secret:
      create: false
      name: "prefect-postgresql-credentials"

    ingress:
      enabled: true
      className: nginx
      host:
        hostname: <http://prefect-reseng-dev.chicagotrading.io|prefect-reseng-dev.chicagotrading.io>
      tls: true
      annotations:
        <http://cert-manager.io/cluster-issuer|cert-manager.io/cluster-issuer>: chicagotradingio
      extraPaths:
      - path: /api
        pathType: Prefix
        backend:
          service:
            name: prefect-server
            port:
              number: 4200

    # Set image registry to use Artifactory
    image:
      registry: <http://artifactory.chicagotrading.com/docker-all|artifactory.chicagotrading.com/docker-all>

---
apiVersion: <http://secrets.hashicorp.com/v1beta1|secrets.hashicorp.com/v1beta1>
kind: VaultStaticSecret
metadata:
  name: prefect-postgresql-credentials
  namespace: prefect-dev
spec:
  destination:
    create: true
    name: prefect-postgresql-credentials
  mount: azure_kv
  path: research_engineering/azure/ncus/psql-admin-dev
  refreshAfter: 30m
  type: kv-v2
  vaultAuthRef: prefect-dev
---
apiVersion: <http://secrets.hashicorp.com/v1beta1|secrets.hashicorp.com/v1beta1>
kind: VaultAuth
metadata:
  name: prefect-dev
  namespace: prefect-dev
spec:
  kubernetes:
    role: prefect-dev
    serviceAccount: prefect-dev
    tokenExpirationSeconds: 600
  method: kubernetes
  mount: kubernetes
m
✅ thought for 103.0 seconds
Thanks for sharing your config. You’ll want to do two things: - Provide a logging.yml (with Prefect’s JsonFormatter) to your pods - Point Prefect at that file via PREFECT_LOGGING_SETTINGS_PATH Because you’re using Helm, the most reliable approach is a ConfigMap + mounts. 1) Create the ConfigMap with your logging.yml Apply this to the prefect-dev namespace alongside your HelmRelease:
Copy code
apiVersion: v1
kind: ConfigMap
metadata:
  name: prefect-logging
  namespace: prefect-dev
data:
  logging.yml: |
    version: 1
    disable_existing_loggers: false
    formatters:
      json:
        '()': prefect.logging.formatters.JsonFormatter
        fmt: default
    handlers:
      console:
        class: prefect.logging.handlers.PrefectConsoleHandler
        stream: <ext://sys.stdout>
        formatter: json
      api:
        class: prefect.logging.handlers.APILogHandler
    loggers:
      prefect:
        level: INFO
        handlers: [console, api]
        propagate: false
      prefect.flow_runs:
        level: INFO
        handlers: [console, api]
        propagate: false
      prefect.task_runs:
        level: INFO
        handlers: [console, api]
        propagate: false
    root:
      level: WARNING
      handlers: [console]
2) Patch your HelmRelease values to mount and use it You already set env for server and backgroundServices. Add these: - For the Prefect Server API/UI/background pods (server + backgroundServices): - Mount the ConfigMap at /etc/prefect/logging - Set PREFECT_LOGGING_SETTINGS_PATH=/etc/prefect/logging/logging.yml Depending on your chart version, the keys are typically extraVolumes/extraVolumeMounts. If your chart exposes those, patch like this:
Copy code
spec:
  values:
    server:
      env:
        - name: PREFECT_LOGGING_SETTINGS_PATH
          value: /etc/prefect/logging/logging.yml
        # keep your existing env here...

      extraVolumes:
        - name: prefect-logging
          configMap:
            name: prefect-logging

      extraVolumeMounts:
        - name: prefect-logging
          mountPath: /etc/prefect/logging
          readOnly: true

    backgroundServices:
      env:
        - name: PREFECT_LOGGING_SETTINGS_PATH
          value: /etc/prefect/logging/logging.yml
        # keep your existing env here...

      extraVolumes:
        - name: prefect-logging
          configMap:
            name: prefect-logging

      extraVolumeMounts:
        - name: prefect-logging
          mountPath: /etc/prefect/logging
          readOnly: true
If your chart uses different keys (e.g., podTemplate, additionalVolumes, additionalVolumeMounts), let me know the chart version and I’ll translate this to the exact fields. 3) Ensure flow-run pods also emit JSON (critical for Datadog) The HelmRelease only affects the server pods. Your flow and task logs come from flow-run pods spawned by your Kubernetes work pool(s). Update the work pool base job template to mount the same ConfigMap and set the env var: - Export the default:
Copy code
prefect work-pool get-default-base-job-template --type kubernetes -f base-template.json
- Edit to add a volume, mount, and env (showing only the additions):
Copy code
{
  "job_configuration": {
    "job_manifest": {
      "spec": {
        "template": {
          "spec": {
            "volumes": [
              {
                "name": "prefect-logging",
                "configMap": {
                  "name": "prefect-logging"
                }
              }
            ],
            "containers": [
              {
                "name": "prefect-job",
                "env": [
                  {
                    "name": "PREFECT_LOGGING_SETTINGS_PATH",
                    "value": "/etc/prefect/logging/logging.yml"
                  }
                ],
                "volumeMounts": [
                  {
                    "name": "prefect-logging",
                    "mountPath": "/etc/prefect/logging",
                    "readOnly": true
                  }
                ]
              }
            ]
          }
        }
      }
    }
  }
}
- Apply it:
Copy code
prefect work-pool update "<your-pool-name>" --base-job-template base-template.json
Notes
- Your namespace annotation suggests you already have cluster log collection; once Prefect logs are JSON on stdout, Datadog will auto-parse them. - If your worker(s) are also deployed via Helm, apply the same mount + env pattern to the worker chart so worker logs are JSON too. If you can share that HelmRelease (chart/version), I’ll give you an exact patch. - Alternative: bake logging.yml into the image at ~/.prefect/logging.yml and skip the mount; then you don’t need PREFECT_LOGGING_SETTINGS_PATH. Refs - Prefect logging customization: Logging customization - Helm charts: prefect-helm - Kubernetes work pools: Kubernetes work pools If you confirm your Helm chart version for prefect-server and prefect-worker, I’ll provide an exact, copy-paste values patch for both.
28 Views
Open in Slack
PreviousNext
Powered by