Run Filebeat on Cloud Foundry

edit

You can use Filebeat on Cloud Foundry to retrieve and ship logs.

Create Cloud Foundry credentials

edit

To connect to loggregator and receive the logs, Filebeat requires credentials created with UAA. The uaac command creates the required credentials for connecting to loggregator.

uaac client add filebeat --name filebeat --secret changeme --authorized_grant_types client_credentials,refresh_token --authorities doppler.firehose,cloud_controller.admin_read_only

Use a unique secret: The uaac command shown here is an example. Remember to replace changeme with your secret, and update the filebeat.yml file to use your chosen secret.

Download Cloud Foundry deploy manifests

edit

You deploy Filebeat as an application with no route.

Cloud Foundry requires that 3 files exist inside of a directory to allow Filebeat to be pushed. The commands below provide the basic steps for getting it up and running.

curl -L -O https://artifacts.elastic.co/downloads/beats/filebeat/filebeat-8.2.3-linux-x86_64.tar.gz
tar xzvf filebeat-8.2.3-linux-x86_64.tar.gz
cd filebeat-8.2.3-linux-x86_64
curl -L -O https://raw.githubusercontent.com/elastic/beats/8.2/deploy/cloudfoundry/filebeat/filebeat.yml
curl -L -O https://raw.githubusercontent.com/elastic/beats/8.2/deploy/cloudfoundry/filebeat/manifest.yml

You need to modify the filebeat.yml file to set the api_address, client_id and client_secret.

Load Kibana dashboards

edit

Filebeat comes packaged with various pre-built Kibana dashboards that you can use to visualize data in Kibana.

If these dashboards are not already loaded into Kibana, you must run the Filebeat setup command. To learn how, see Load Kibana dashboards.

The setup command does not load the ingest pipelines used to parse log lines. By default, ingest pipelines are set up automatically the first time you run Filebeat and connect to Elasticsearch.

If you are using a different output other than Elasticsearch, such as Logstash, you need to:

Deploy Filebeat

edit

To deploy Filebeat to Cloud Foundry, run:

cf push

To check the status, run:

$ cf apps

name       requested state   instances   memory   disk   urls
filebeat   started           1/1         512M     1G

Log events should start flowing to Elasticsearch. The events are annotated with metadata added by the add_cloudfoundry_metadata processor.

Scale Filebeat

edit

A single instance of Filebeat can ship more than a hundred thousand events per minute. If your Cloud Foundry deployment is producing more events than Filebeat can collect and ship, the Firehose will start dropping events, and it will mark Filebeat as a slow consumer. If the problems persist, Filebeat may be disconnected from the Firehose. In such cases, you will need to scale Filebeat to avoid losing events.

The main settings you need to take into account are:

  • The shard_id specified in the cloudfoundry input configuration. The Firehose will divide the events amongst all the Filebeat instances with the same value for this setting. All the instances with the same shard_id should have the same configuration.
  • Number of Filebeat instances. When Filebeat is deployed as a Cloud Foundry application, it can be scaled up and down like any other application, with cf scale or by specifying the number of instances in the manifest.
  • Output configuration. In some cases, you can fine-tune the output configuration to improve the events throughput. Some outputs support multiple workers. The number of workers can be changed to take better advantage of the available resources.

Some basic recommendations to adjust these settings when Filebeat is not able to collect all events:

  • If Filebeat is hitting its CPU limits, you will need to increase the number of Filebeat instances deployed with the same shard_id.
  • If Filebeat has some spare CPU, there may be some backpressure from the output. Try to increase the number of workers in the output. If this doesn’t help, the bottleneck may be in the network or in the service receiving the events sent by Filebeat.
  • If you need to modify the memory limit of Filebeat, remember that CPU shares assigned to Cloud Foundry applications depend on the configured memory limit. You may need to check the other recommendations after that.