Snowflake integration

Run the following command in your CLI to clone the New Relic Snowflake repo:

$
git clone https://github.com/newrelic/newrelic-snowflake-integration.git

From the releases section, download the snowflakeintegration-linux binary. Place that binary in the cloned repo.

Make sure the binary is executable by running:

$
chmod +x snowflakeintegration-linux

To ensure you're using the most recent information for the integration, update your config.yaml file so that the file matches the snippet found in Snowflake's Key Pair authentication section.

If the New Relic infrastructure agent runs as a systemd service, follow these steps to set the environment variable. (Skip this step if you're not running the New Relic infrastructure agent as a systemd service and follow the Step 5).

• To determine if you're on a system using systemd as the init service, run the following:

$
[[ `\systemctl` =~ -\.mount ]] && echo yes || echo no

Set the NEWRELIC_SNOWFLAKE_HOME environment variable as documented in the Setting NEWRELIC_SNOWFLAKE_HOME section. (Skip this step if you're running the New Relic infrastructure agent as a systemd service and follow the Step 4).

Copy the relevant flex config for your platform from flexConfigs to the agent's integrations.d folder:

• for Linux, find it at /etc/newrelic-infra/integrations.d/
• for Windows, find it at C:\Program Files\New Relic\newrelic-infra\integrations.d\

To find the queries that take the longest to load:

1. Create a file and name it longest_queries.sql

2. Designate the file path as newrelic-snowflake-integrations/queries.

3. In longest_queries.sql, add the following snippet:

   select 
       query_id, 
       query_text,
       (execution_time / 60000) as exec_time,
       warehouse_name,
       user_name,
       execution_status
   from 
       snowflake.account_usage.query_history 
   where 
       execution_status = 'SUCCESS' 
   order by 
       execution_time desc

   Copy

4. In the flex-snowflake-linux.yml file, add the following snippet:

- name: longestQueries
entity: snowflake
# New Relic will capture all your Snowflake metrics when you use `event_type: SnowflakeVirtualWarehouse`. 
event_type: SnowflakeVirtualWarehouse
custom_attributes:
    metric_type: snowflake.query_performance
commands: 
    - run: YOUR_PATH_TO_DOWNLOADED_BINARY_FILE YOUR_PATH_TO_CLONED_REPOSITORY_DIRECTORY/config.yaml YOUR_PATH_TO_CLONED_REPOSITORY_DIRECTORY/queries/longest_queries.sql

To find the queries that take the longest to load:

1. Create a file and name it credit_usage_by_warehouse.sql

2. Designate the file path as YOUR_PATH_TO_CLONED_REPOSITORY_DIRECTORY/queries/credit_usage_by_warehouse.sql.

3. In credit_usage_by_warehouse.sql, add the following snippet:

   select 
       warehouse_name, 
       sum(credits_used) as total_credits_used 
   from 
       snowflake.account_usage.warehouse_metering_history 
   group by 
       1 
   order by 
       2 desc;

   Copy

4. Add the following snippet to flex-snowflake.yml:

   - name: CreditUsageByWarehouse
           entity: snowflake
           event_type: SnowflakeVirtualWarehouse
           custom_attributes:
               metric_type: snowflake.credit_usage_by_warehouse
           commands: 
           - run: YOUR_PATH_TO_DOWNLOADED_BINARY_FILE YOUR_PATH_TO_CLONED_REPOSITORY_DIRECTORY/config.yaml YOUR_PATH_TO_CLONED_REPOSITORY_DIRECTORY/queries/credit_usage_by_warehouse.sql

   Copy