Skip to main content

Best practices

Estimate before a full run​

Always run apibackuper estimate full first. It reports record counts, expected size, and wall-clock time so you can set rate limits and storage accordingly.

Prefer YAML and keep secrets out of git​

Use apibackuper.yaml. Put tokens and passwords in files referenced by token_file / password_file, not in the committed config:

auth:
  type: bearer
  token_file: /run/secrets/api_token

Checkpoint long jobs​

settings:
  checkpoint_file: apibackuper_checkpoint.json
  checkpoint_interval_pages: 5
  state_file: apibackuper_state.json

Resume with:

apibackuper run full --resume

Respect API limits​

Set both rate_limit and request.parallelism. Concurrency still waits on the rate limiter. Start conservative (parallelism: 1) and raise it only after a successful estimate.

Identify records uniquely​

Set data.item_key (comma-separated for composite keys) so incremental and update modes can skip duplicates. Set data.change_key for update mode.

Validate configuration​

apibackuper validate-config

Do this after every config edit, especially pagination and auth changes.

Keep SSL enabled​

Leave request.verify_ssl: true unless the API has a known certificate problem. The tool logs a warning when verification is off.