Fabric Cleanup Processes

This section provides information on how to configure the janitor maintenance utilities built into the Fabric Helm chart.

Fabric Janitor

The Fabric Helm chart includes a Janitor section that is a collection of jobs and utilities for managing long-running Fabric deployments. The Janitor processes can be enabled and configured by the following section in the values.yaml:

# Janitor processes for cortex
janitor:
  # janitor.enabled: enable/disable all janitor processes/cronjobs/etc.
  enabled: true
  deployment:
    annotations:
      sidecar.istio.io/inject: "false"
  persistence:
    # janitor.persistence.size: the size of the PVC to request for performing archives/backups (should be consistent with the size of other PVC's)
    size: 100Gi
  # janitor.synapse: settings related to SynapseState/Activation Record archive and backup from mongodb
  synapse:
    # janitor.synapse.archive: Enable/disable archiving of SynapseState collection within the configured object store's managed content bucket
    archive: false
    # janitor.synapse.image: the docker image used for performing the archive and cleanup of mongodb records
    image: mongo:3.6
    # janitor.synapse.retentionDays: the number of days to retain activation records within the database, any older records will be archived into S3/managed content and removed from the database
    retentionDays: 30
    # janitor.synapse.schedule: the crontab schedule used for running the synapse cleanup CronJob, default is daily (at midnight)
    schedule: "0 0 * * *" # Runs daily

Synapse Janitor Process

The Synapse Janitor process is a cron job that runs on an interval (janitor.synapse.schedule) and removes any Agent activations that are older than janitor.synapse.retentionDays. Optionally, the Synapse Janitor can also create an archive of the activations and upload them to the configured Managed Content bucket prior to deleting the activation records.

NOTE

Archiving to Managed Content currently does not work in clusters with IRSA enabled.