Logging

Structured logs on stderr

The daemon and the dry-run path emit structured logs through tracing, written to stderr. Stdout is reserved for machine-readable dry-run JSON, so you can pipe a plan without log lines polluting it.

Control verbosity with the RUST_LOG environment variable:

RUST_LOG=debug proton-syncd \
  --local-root /tmp/proton-sync-demo \
  --remote-root /Drive/RemoteFolder

Useful values are error, warn, info, and debug. When RUST_LOG is unset, the daemon defaults to info.

Retry attempts, unsuccessful proton-drive CLI exits, and command timeouts are logged with structured fields — operation, attempt, exit status, stderr, and timeout milliseconds — so you can see exactly which remote call misbehaved.

Under systemd, these logs go to the journal:

journalctl --user -u proton-syncd.service -f

Status and metrics sidecars

Alongside logs, the daemon persists two JSON files next to its SQLite index, for local inspection by scripts, service monitors, and the desktop app. They are written atomically and are always ignored by sync.

File (for sync_index.db)	Contents
sync_index.status.json	Recent status history — the last 20 entries: timestamp, message, last error, plan summary, and successful-sync summary. Reloaded on restart, so recent failures and successes survive a restart.
sync_index.metrics.json	A metrics snapshot for operational inspection.

These sidecars — not the live SQLite index — are the safe way to read daemon state, because the index runs without WAL and a reader would race the writer. The desktop app and the proton-sync history command both surface the status history.

See the CLI reference for proton-sync status and proton-sync history, which expose the same information over the control socket.