Export Jellyfin playback statistics to Prometheus and Grafana. See https://sno.ws/jellyfin-stats for more info

json-exporter-config.yaml

modules:
  jellyfin:
    headers:
      # The Token value here needs to be an API key generated from the
      # Jellyfin admin panel. It's hard-coded here but I'm sure there's
      # a better way
      Authorization: MediaBrowser Token=ADD_TOKEN_HERE
      Content-Type: application/json
      accept: application/json

    # This will return all active sessions regardless of
    # whether something is playing. You can use a combination
    # of label and value filters in Grafana to only get actively
    # playing sessions
    metrics:
    - name: jellyfin
      type: object
      help: User playback metrics from Jellyfin
      # Only look at sessions with a NowPlayingItem key, per @abe-kanofsky
      path: '{ [?(@.NowPlayingItem)] }'
      labels:
        user_name: '{ .UserName }'
        # User PromQL label_join and label_replace to concatenate
        # these values into a nice item description
        item_type: '{ .NowPlayingItem.Type }'
        item_name: '{ .NowPlayingItem.Name }'
        item_path: '{ .NowPlayingItem.Path }'
        series_name: '{ .NowPlayingItem.SeriesName }'
        episode_index: 'e{ .NowPlayingItem.IndexNumber }'
        season_index: 's{ .NowPlayingItem.ParentIndexNumber }'
        client_name: '{ .Client }'
        device_name: '{ .DeviceName }'
        # Include the unique session ID in case the above info isn't unique
        session_id: '{ .Id }'
      values:
        is_paused: '{ .PlayState.IsPaused }'

prometheus-config.yaml

scrape_configs:
  - job_name: json
    metrics_path: /probe
    params:
      # The name of the module defined by json-exporter-config.yaml
      module: [jellyfin] 
    static_configs:
      - targets:
        # Use the Sessions endpoint to see actively playing items
        - https://jellyfin.example.com/Sessions
    relabel_configs:
      - source_labels: [__address__]
        target_label: __param_target
      - source_labels: [__param_target]
        target_label: instance
      - target_label: __address__
        replacement: HOSTNAME:9115 # The exporter's hostname:port

state-timeline-grafana-panel.json

{
  "datasource": {
    "type": "prometheus",
    "uid": "D5GQA944k"
  },
  "description": "",
  "fieldConfig": {
    "defaults": {
      "custom": {
        "lineWidth": 0,
        "fillOpacity": 70,
        "spanNulls": false,
        "insertNulls": false,
        "hideFrom": {
          "tooltip": false,
          "viz": false,
          "legend": false
        }
      },
      "color": {
        "mode": "continuous-GrYlRd"
      },
      "mappings": [],
      "thresholds": {
        "mode": "absolute",
        "steps": [
          {
            "color": "green",
            "value": null
          },
          {
            "color": "red",
            "value": 80
          }
        ]
      }
    },
    "overrides": []
  },
  "gridPos": {
    "h": 6,
    "w": 13,
    "x": 0,
    "y": 9
  },
  "id": 348,
  "options": {
    "mergeValues": true,
    "showValue": "auto",
    "alignValue": "left",
    "rowHeight": 0.9,
    "legend": {
      "showLegend": true,
      "displayMode": "list",
      "placement": "bottom"
    },
    "tooltip": {
      "mode": "single",
      "sort": "none"
    }
  },
  "pluginVersion": "10.1.1",
  "targets": [
    {
      "datasource": {
        "type": "prometheus",
        "uid": "D5GQA944k"
      },
      "editorMode": "code",
      "expr": "sum by (user_name, item_index_clean, item_type) (label_replace(label_join(label_join(jellyfin_is_paused{job=\"json\"}, \"item_index\", \"\", \"season_index\", \"episode_index\"), \"full_name\", \" - \", \"series_name\", \"item_index\", \"item_name\"), \"item_index_clean\", \"$1\", \"full_name\", \"^[- ]*(.*?)[- ]*$\"))",
      "instant": false,
      "legendFormat": "__auto",
      "range": true,
      "refId": "A"
    }
  ],
  "title": "Jellyfin",
  "transformations": [
    {
      "id": "seriesToRows",
      "options": {}
    },
    {
      "id": "extractFields",
      "options": {
        "keepTime": false,
        "replace": false,
        "source": "Metric"
      }
    },
    {
      "id": "filterByValue",
      "options": {
        "filters": [
          {
            "config": {
              "id": "isNotNull",
              "options": {}
            },
            "fieldName": "user_name"
          }
        ],
        "match": "any",
        "type": "include"
      }
    },
    {
      "id": "calculateField",
      "options": {
        "alias": "Item",
        "mode": "reduceRow",
        "reduce": {
          "include": [
            "item_index_clean"
          ],
          "reducer": "firstNotNull"
        }
      }
    },
    {
      "id": "calculateField",
      "options": {
        "alias": "Name",
        "mode": "reduceRow",
        "reduce": {
          "include": [
            "user_name"
          ],
          "reducer": "firstNotNull"
        }
      }
    },
    {
      "id": "calculateField",
      "options": {
        "alias": "Temp",
        "mode": "reduceRow",
        "reduce": {
          "include": [
            "Item"
          ],
          "reducer": "firstNotNull"
        },
        "replaceFields": false
      }
    },
    {
      "id": "filterByValue",
      "options": {
        "filters": [
          {
            "config": {
              "id": "regex",
              "options": {
                "value": "(Movie|Episode)"
              }
            },
            "fieldName": "item_type"
          }
        ],
        "match": "any",
        "type": "include"
      }
    },
    {
      "id": "filterByValue",
      "options": {
        "filters": [
          {
            "config": {
              "id": "notEqual",
              "options": {
                "value": 1
              }
            },
            "fieldName": "Value"
          }
        ],
        "match": "any",
        "type": "include"
      }
    },
    {
      "id": "partitionByValues",
      "options": {
        "fields": [
          "Name"
        ]
      }
    },
    {
      "id": "filterFieldsByName",
      "options": {
        "include": {
          "pattern": "Temp|Time (.*)"
        }
      }
    },
    {
      "id": "renameByRegex",
      "options": {
        "regex": "Temp (.*)",
        "renamePattern": "$1"
      }
    }
  ],
  "type": "state-timeline"
}

This is super useful stuff!

Just wanna mention an issue I ran into with this setup. It seems that it's possible for Jellyfin to report a user as having more than one session with the same client, device, and NowPlayingItem. When this happens, the JSON exporter doesn't know what to do because it has no way to differentiate the two session JSON objects given the labels it's configured to look at. This causes all requests to the probe endpoint to return an error like this:

An error has occurred while serving metrics:

collected metric "jellyfin_is_paused" { label:{name:"client_name" value:"Jellyfin Web"} label:{name:"device_name" value:"Chrome"} label:{name:"episode_index" value:""} label:{name:"item_name" value:""} label:{name:"item_path" value:""} label:{name:"item_type" value:""} label:{name:"season_index" value:""} label:{name:"series_name" value:""} label:{name:"user_name" value:"..."} untyped:{value:0}} was collected before with the same name and label values

I fixed this by adding session_id: '{ .Id }' to the modules.jellyfin.metrics[0].labels object in the JSON exporter config which is enough to allow it to defferentiate between the different sessions, since Id is always unique even when the user, client, device, and NowPlayingItem are the same. After making this change, I resarted the exporter and it worked normally again.

Another thing: modules.jellyfin.metrics[0].path should be '{ [?(@.NowPlayingItem)] } to look at only sessions that actually have a NowPlayingItem key. Otherwise, the exporter also tries to read sessions without that key and spits out a pretty verbose error each time it encounters one, causing the log file to blow up.

Nice @abe-kanofsky, these are excellent suggestions! I'll update the gist and blogpost to reflect them.