---
title: "Monitor GEM usage | Grafana Enterprise Metrics documentation"
description: "Monitor GEM usage by access policy and token Monitor usage based on access policy and token to gain insight into which users in a tenant are running expensive queries in GEM. GEM reports these usage details in the query-frontend component’s query stats log."
---

# Monitor GEM usage by access policy and token

Monitor usage based on access policy and token to gain insight into which users in a tenant are running expensive queries in GEM. GEM reports these usage details in the query-frontend component’s query stats log.

## Set up usage details logging

> Note
> 
> If you are deploying GEM using the mimir-distributed Helm chart version 5.5 or later, usage details are logged by default.

Set up usage details logging to capture a user’s access policy and token when they run a query in GEM.

1. Use the following CLI flags to send request headers containing a query author’s corresponding access policy and token to the downstream query services:
   
   - `header_x_access_policy_name`: The query author’s access policy
   - `header_x_token_name`: The query author’s token
2. Use the `query-frontend.log-query-request-headers` setting in the configuration file to capture this information in the query stats log.

## View usage details

To view the log for a GEM query, run the following LogQL query:

logql ![Copy code to clipboard](/media/images/icons/icon-copy-small-2.svg) Copy

```logql
{namespace="_`<NAMESPACE>`_", cluster="_`<CLUSTER>`_", container="query-frontend"}
|= `query stats`
```

The returned log file includes the `header_x_access_policy_name` and `header_x_token_name` headers.

You can use this information to run additional queries that aggregate the usage statistics for a specified access policy or token. For example, the following LogQL query measures the time that queries spent in-queue based on the query author’s token:

logql ![Copy code to clipboard](/media/images/icons/icon-copy-small-2.svg) Copy

```logql
avg by (header_x_token_name) (
  avg_over_time(
    {namespace="_`<NAMESPACE>`_", cluster="_`<CLUSTER>`_", container="query-frontend"}
     |= `query stats`
     | logfmt
     | unwrap queue_time_seconds [$__auto]
))
```