Pruning

The sequencer, participant and CometBFT have pruning options that can be used to ensure storage use is kept within reasonable bounds.

Pruning can also be configured for the participant, but we don’t currently recommend enabling it for the SVs.

Sequencer pruning

Can be enabled in helm through the following config:

domain:
  sequencerPruningConfig:
    # Enable or disable sequencer pruning
    enabled: true
    # The pruning interval is the time between two consecutive prunings.
    pruningInterval: "1 hour"
    # The retention period is the time for which the sequencer will retain the data.
    retentionPeriod: "30 days"

Note

It is recommended that sequencer pruning is enabled and the pruningInterval is set to 1 hour and the retentionPeriod to 30 days.

CometBFT pruning

It is enabled by default. Pruning is defined as the number of blocks to keep. Older blocks are pruned.

The number of blocks to keep can be configured under the node helm values key.

  # Number of blocks to keep, used for pruning. 0 -> keep all blocks.
  # Number of blocks to keep for 30 days with an upper bound of 7k blocks/h.
  retainBlocks: 5040000

Participant pruning

Participant pruning is also supported and recommend. To enable it, set the following helm value on your validator chart:

participantPruningSchedule:
  cron: 0 /10 * * * ? # Run every 10min
  maxDuration: 5m # Run for a max of 5min per iteration
  retention: 30d # Retain history that is newer than 30d.

You also need to tell the participant to continue pruning even if it has not received an ACS commitment from one of its counter participant in the last 30 days. Without this pruning will essentially never run on mainnet as validators get shut down relatively frequently: To do so, set the following through the additionalEnvVars on your participant:

additionalEnvVars:
    - name: ADDITIONAL_CONFIG_PRUNING_ACS_COMITMENTS
      value: |
        canton.participants.participant.parameters.stores.safe-to-prune-commitment-state = "all"

Comments