yuecong / azure-metrics-exporter

Azure Insights metrics exporter for Prometheus

Azure Insights metrics exporter

Prometheus exporter for Azure Insights metrics (on demand). Supports metrics fetching from all resource with one scrape (automatic service discovery) and also supports dimensions.

Configuration (except Azure connection) of this exporter is made entirely in Prometheus instead of a seperate configuration file, see examples below.

Configuration

Normally no configuration is needed but can be customized using environment variables.

Usage:
  azure-metrics-exporter [OPTIONS]

Application Options:
      --debug                              debug mode [$DEBUG]
  -v, --verbose                            verbose mode [$VERBOSE]
      --log.json                           Switch log output to json format [$LOG_JSON]
      --azure-environment=                 Azure environment name (default: AZUREPUBLICCLOUD) [$AZURE_ENVIRONMENT]
      --concurrency.subscription=          Concurrent subscription fetches (default: 5) [$CONCURRENCY_SUBSCRIPTION]
      --concurrency.subscription.resource= Concurrent requests per resource (inside subscription requests) (default:
                                           10) [$CONCURRENCY_SUBSCRIPTION_RESOURCE]
      --enable-caching                     Enable internal caching [$ENABLE_CACHING]
      --bind=                              Server address (default: :8080) [$SERVER_BIND]

Help Options:
  -h, --help                               Show this help message

for Azure API authentication (using ENV vars) see https://github.com/Azure/azure-sdk-for-go#authentication

Metrics

Metric	Description
azurerm_stats_metric_collecttime	General exporter stats
azurerm_stats_metric_requests	Counter of resource metric requests with result (error, success)
azurerm_resource_metric	Resource metrics exported by probes (can be changed using name parameter)
azurerm_loganalytics_query_result	LogAnalytics rows exported by probes

HTTP Endpoints

Endpoint	Description
/metrics	Default prometheus golang metrics
/probe/metrics/resource	Probe metrics for one resource (see azurerm_resource_metric)
/probe/metrics/list	Probe metrics for list of resources (see azurerm_resource_metric)
/probe/metrics/scrape	Probe metrics for list of resources and config on resource by tag name (see azurerm_resource_metric)
/probe/loganalytics/query	Probe metrics from LogAnalytics query (see azurerm_loganalytics_query_result)

/probe/metrics/resource parameters

GET parameter	Default	Required	Multiple	Description
subscription		yes	yes	Azure Subscription ID
target		yes	yes	Azure Resource URI
timespan	PT1M	no	no	Metric timespan
interval		no	no	Metric timespan
metric		no	yes	Metric name
aggregation		no	yes	Metric aggregation (minimum, maximum, average, total, count, multiple possible separated with ,)
name	azurerm_resource_metric	no	no	Prometheus metric name
metricFilter		no	no	Prometheus metric filter (dimension support)
metricTop		no	no	Prometheus metric dimension count (dimension support)
metricOrderBy		no	no	Prometheus metric order by (dimension support)
cache	(same as timespan)	no	no	Use of internal metrics caching

Hint: Multiple values can be specified multiple times or with a comma in a single value.

/probe/metrics/list parameters

GET parameter	Default	Required	Multiple	Description
subscription		yes	yes	Azure Subscription ID (or multiple separate by comma)
filter		yes	no	Azure Resource filter (https://docs.microsoft.com/en-us/rest/api/resources/resources/list)
timespan	PT1M	no	no	Metric timespan
interval		no	no	Metric timespan
metric		no	yes	Metric name
aggregation		no	yes	Metric aggregation (minimum, maximum, average, total, count, multiple possible separated with ,)
name	azurerm_resource_metric	no	no	Prometheus metric name
metricFilter		no	no	Prometheus metric filter (dimension support)
metricTop		no	no	Prometheus metric dimension count (dimension support)
metricOrderBy		no	no	Prometheus metric order by (dimension support)
cache	(same as timespan)	no	no	Use of internal metrics caching

Hint: Multiple values can be specified multiple times or with a comma in a single value.

/probe/metrics/scrape parameters

GET parameter	Default	Required	Multiple	Description
subscription		yes	yes	Azure Subscription ID (or multiple separate by comma)
filter		yes	no	Azure Resource filter (https://docs.microsoft.com/en-us/rest/api/resources/resources/list)
metricTagName		yes	no	Resource tag name for getting "metric" list
aggregationTagName		yes	no	Resource tag name for getting "aggregation" list
timespan	PT1M	no	no	Metric timespan
interval		no	no	Metric timespan
metric		no	yes	Metric name
aggregation		no	yes	Metric aggregation (minimum, maximum, average, total, multiple possible separated with ,)
name	azurerm_resource_metric	no	no	Prometheus metric name
metricFilter		no	no	Prometheus metric filter (dimension support)
metricTop		no	no	Prometheus metric dimension count (integer, dimension support)
metricOrderBy		no	no	Prometheus metric order by (dimension support)
cache	(same as timespan)	no	no	Use of internal metrics caching

Hint: Multiple values can be specified multiple times or with a comma in a single value.

/probe/loganalytics/query parameters

GET parameter	Default	Required	Description
workspace		yes	Azure LogAnalytics workspace ID
query		yes	LogAnalytics query
timespan		yes	Query timespan

Prometheus configuration

Azure Redis metrics

- job_name: azure-metrics-redis
  scrape_interval: 1m
  metrics_path: /probe/metrics/list
  params:
    name: ["my_own_metric_name"]
    subscription: 
    - xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
    filter: ["resourceType eq 'Microsoft.Cache/Redis'"]
    metric:       
    - connectedclients
    - totalcommandsprocessed
    - cachehits
    - cachemisses
    - getcommands
    - setcommands
    - operationsPerSecond
    - evictedkeys
    - totalkeys
    - expiredkeys
    - usedmemory
    - usedmemorypercentage
    - usedmemoryRss
    - serverLoad
    - cacheWrite
    - cacheRead
    - percentProcessorTime
    - cacheLatency
    - errors
    interval: ["PT1M"]
    timespan: ["PT1M"]
    aggregation:  
    - average
    - total
  static_configs:
  - targets: ["azure-metrics:8080"]

Virtual Gateway metrics

- job_name: azure-metrics-virtualNetworkGateways
  scrape_interval: 1m
  metrics_path: /probe/metrics/list
  params:
    name: ["my_own_metric_name"]
    subscription: 
    - xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
    filter: ["resourceType eq 'Microsoft.Network/virtualNetworkGateways'"]
    metric:       
    - AverageBandwidth
    - P2SBandwidth
    - P2SConnectionCount
    - TunnelAverageBandwidth
    - TunnelEgressBytes
    - TunnelIngressBytes
    - TunnelEgressPackets
    - TunnelIngressPackets
    - TunnelEgressPacketDropTSMismatch
    - TunnelIngressPacketDropTSMismatch
    interval: ["PT5M"]
    timespan: ["PT5M"]
    aggregation:
    - average
    - total
  static_configs:
  - targets: ["azure-metrics:8080"]

Virtual Gateway connection metrics (dimension support)

- job_name: azure-metrics-virtualNetworkGateways-connections
  scrape_interval: 1m
  metrics_path: /probe/metrics/list
  params:
    name: ["my_own_metric_name"]
    subscription:
    - xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
    filter: ["resourceType eq 'Microsoft.Network/virtualNetworkGateways'"]
    metric:
    - TunnelAverageBandwidth
    - TunnelEgressBytes
    - TunnelIngressBytes
    - TunnelEgressPackets
    - TunnelIngressPackets
    - TunnelEgressPacketDropTSMismatch
    - TunnelIngressPacketDropTSMismatch
    interval: ["PT5M"]
    timespan: ["PT5M"]
    aggregation:  
    - average
    - total
    # by connection (dimension support)
    metricFilter: ["ConnectionName eq '*'"]
    metricTop: ["10"]
  static_configs:
  - targets: ["azure-metrics:8080"]

In these examples all metrics are published with metric name my_own_metric_name.

The List of supported metrics is available in the Microsoft Azure docs.