Scaling

info

Looking for the GitGuardian legacy application Scaling page? Please visit the Scaling (legacy) page.

For information on the new application, as well as determining whether you are using the New or Legacy GitGuardian application, explore the New GitGuardian Architecture page.

Application topology

GitGuardian application consists of several Kubernetes resources.

Helm-based installation facilitates the configuration of all deployments, while also offering enhanced customization options for the following:

• Creation of new classes of workers
• Customization of other resource types such as ephemeral storage, huge pages, etc.
• Provision of nodeSelector, toleration, and more

Kind	Deployment Name	Usage
Front	nginx	Dashboard Frontend and proxy for backend
Backend	app_exporter	(optional) Open Metrics Exporter for applicative metrics
Backend	hook	VCS webhooks events receiver
Backend	internal_api	Backend for the Dashboard (previously gitguardian-app in legacy)
Backend	internal_api_long	Backend for the Dashboard (no timeout)
Backend	public_api	Public API and GGshield scans
Scheduler	beat	Celery Beat task scheduler
Worker	email	Workers for queues: email, notifier
Worker	long	Workers for long tasks: check/install health, asynchronous cleanup tasks, ...
Worker	scanners	Workers for historical scans
Worker	worker	Workers for queues: celery (default), check_run, realtime, realtime_retry
Job	pre-deploy	Pre-deployment job performing database migrations
Job	post-deploy	Post-deployment job performing long data migrations

Configure scaling settings

Each deployment can be configured using the replicas property. For web pods, you will use the webapps.[name].replicas property, and for async workers, you will use the celeryWorkers.[name].replicas.

You can also configure resources requests and limits.

Example

migration:
  # Set resources for pre-deploy and post-deploy jobs
  resources:
    limits:
      cpu: 1000m
      memory: 500Mi
front:
  nginx:
    # Set resources for nginx init containers
    init:
      resources:
        limits:
          cpu: 1000m
          memory: 500Mi
    replicas: 2
    resources:
      limits:
        memory: 1Gi
webapps:
  public_api:
    replicas: 5
    resources:
      requests:
        cpu: 200m
        memory: 500Mi
      limits:
        memory: 4Gi
celeryWorkers:
  scanner:
    replicas: 8
    resources:
      requests:
        cpu: 200m
        memory: 4Gi
      limits:
        memory: 16Gi

Scaling for historical scans of repositories up to 15GB in size

Performing your first Historical Scans

When you add a high number of sources, consider temporarily increasing the number pods for the time of the initial historical scan. Afterward, you can decrease those pods' replicas and resources.

When performing a historical scan, GitGuardian clones the git repository on the pod's ephemeral storage and will traverse all branches and commits in search of potential Policy Breaks. The full scan is done by a single Pod and can last from a few seconds to many hours depending on the repository size.

The more pods you'll add, the more historical scans can be done concurrently. When sizing your nodes, keep in mind that each Pod must have enough ephemeral storage and memory to run.

The following sizing has been tested for 7000 repositories up to 15GB with 16 pods:

Component	Required Capacity	Count
Compute nodes	8 vCPU 64 GB RAM 50GB ephemeral disk space, 500 GB persistent disk space	6
PostgreSQL Master	16 vCPU 64 GB memory 300 GB disk space	1
PostgreSQL Read Replica	8 vCPU 32 GB memory 300 GB disk space	1
scanner Pods	Memory request and limit: 16GB	16

Additional tuning

On Helm-based installations, additional configuration of scanner workers can bring more performance and stability.

In the following example, we specify that scanner workers only use "On Demand" VMs with nvme disks and that pods' ephemeral storage will use these disks

celeryWorkers:
  scanners:
    replicas: 16
    localStoragePath: /nvme/disk # Used for pods ephemeral storage
    nodeSelector: # Must run on "On Demand" nodes with nvme disks
      eks.amazonaws.com/capacityType: ON_DEMAND
      local-nvme-ready: 'true'
    tolerations:
      - key: worker-highdisk
        operator: Equal
        value: 'true'
        effect: NoSchedule
    resources:
      requests:
        cpu: 200m
        memory: 16Gi
      limits:
        memory: 16Gi

Scaling real-time scans

Real-time scans are triggered by Push events sent by the VCS to GitGuardian. Those scan duration is often under a second and should always be under 3 seconds. But to be able to handle peaks of pushes, you may want to increase the count of worker Pods that are processing real-time scans.

We successfully tested peaks of 2000 pushes per hour with 8 worker Pods replicas, without changing default resources settings.

Was this page helpful?