SV Pruning

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

Sequencer pruning

Can be enabled in helm through the following config:

synchronizers:
  current:
    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"
    # Set to true when using CantonBFT
    enableBftSequencer: false

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

The retentionPeriod limits how far behind a node can fall. When the SV app prunes, it disables any other node on your sequencer, such as a validator’s participant, that has not acknowledged events from within the retentionPeriod, and a disabled node cannot reconnect to your sequencer. After a logical synchronizer upgrade, this includes validators that have not yet connected to the successor synchronizer.

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

Overview

Validator Deployment

Super Validator Deployment

Splice Fundamentals

Canton Console

Production Operations

Dedicated Synchronizers

Troubleshooting

Release Notes

Reference

Community Resources

Sequencer pruning

CometBFT pruning

Participant pruning

​Sequencer pruning

​CometBFT pruning

​Participant pruning

Sequencer pruning

CometBFT pruning

Participant pruning