Web UI

dccd ships a small web UI — a dashboard to watch collection, browse stored data, run backfills and control streams without touching the CLI. It is a pure client of the HTTP API: every action is just an API call, so nothing is hidden from scripting.

Running it

dccd ui                     # UI + API only (no scheduler)
dccd start                  # full daemon: scheduler + streams + UI

dccd ui --host 0.0.0.0 --port 8080

Both read settings.ui_host / ui_port from the config. Requires the extra: pip install "dccd[daemon,ui]". Then open http://127.0.0.1:8080.

To protect it, set settings.ui_auth_token — the API then requires a bearer token and the UI injects it automatically. For untrusted networks, keep the default 127.0.0.1 bind and/or put it behind a reverse proxy.

Navigation

A single top bar carries the brand (logo · dccd · version) on the left and the navigation on the right. The nav groups pages by purpose: Dashboard and Data are top-level, while Collect ▾ (Historical, Live) and System ▾ (Logs, Config, Storage) are drop-down menus.

Dates are shown in the timezone set by settings.timezone (local by default; also UTC or any zoneinfo name such as Europe/Paris). Relative ages (“5 min ago”) are timezone-independent.

Pages

Page	What it shows
Dashboard	A KPI bar (datasets, rows stored, on-disk size, live streams, runs in progress), an Active now panel (running backfills with a live progress bar + live streams), recent runs, and a per-exchange data summary.
Data	Read-only view of every dataset on disk under data_path — data-type tabs (OHLC / Trades / Order Book) → per-exchange accordions. Each row shows rows, time range, freshness dot, OHLC gap %, on-disk size, and file count. (Formerly Inventory; /inventory still redirects here.)
Historical	Manage backfill jobs (OHLC and Trades): data-type tabs → per-exchange accordions → one row per dataset with an editable first date, a Schedule (Off / hourly / daily / custom — a recurring backfill run by the daemon), a real coverage bar (first date → today, reflecting stored data and holes), and inline Run / Delete. Run all (page header) and per-exchange Run all trigger every job at once. Order books have no REST history, so they live only on Live.
Live	Manage streams (Trades and Order Book): same tabs/accordions, with a liveness indicator (last price/quote + freshness) per stream and inline Start / Stop / Delete. Order-book streams expose a snapshot every N s interval. (OHLC is collected on Historical via a Schedule, not streamed live.)
Config	Edit settings / storage / alerts as a form (or the whole config as raw JSON) and save back to config.yml — validated server-side. Jobs are managed on Historical / Live (or the raw-JSON tab for bulk edits).
Logs	Recent runs first (each with a human label and an expandable log tail), plus a live SSE console of whatever is running right now.
Storage	On-disk dataset breakdown by exchange with sizes and file counts.

Running backfills & watching streams

On Historical, set a row’s first date and click Run to fill the missing history. The coverage bar turns live and updates by time covered (and the row count) as the run progresses; Run becomes Stop, which cancels cooperatively and keeps everything already collected. The default last start uses a bounded look-back (short for trades) so a first click can’t trigger a runaway download.

On Live, the liveness column proves a stream is actually receiving data: a coloured dot (green while data is arriving within the data’s cadence, then amber, then grey), the last value, and how long ago. It is seeded from the last on-disk point, so a page refresh shows freshness immediately rather than “waiting…”; the age reads as a relative “N min ago” for the last day and an absolute date beyond, or the last-run date-time once a stream is stopped.

Note

Data lists all data on disk, independent of your configured jobs (e.g. imported or migrated history). On Historical, datasets on disk that aren’t tracked by a job are listed separately with a + Track button to adopt them as a job.