Broker connection errors

Broker problems cluster into a handful of shapes. Start from the symptom:

401 unauthorized: broker rejected your keys

The keys are wrong: a paste typo, a pair revoked at Alpaca, or paper keys in the live slot (and vice versa; Alpaca’s paper and live systems each reject the other’s keys). Regenerate at Alpaca, re-paste both halves into the matching environment card, and save. Saving re-runs the automatic connection test against Alpaca before anything is stored.

429 rate-limited: too many requests too fast

Too many requests in a window, usually transient. A scheduled background sync that hits a 429 is automatically retried a few times before it gives up, so a brief spike clears on its own. A page you are looking at does not retry a 429 — the limit is counted per minute, so trying again immediately would only add to it — and reloading a moment later is the right move. If you see it persistently, something else is hammering the same Alpaca account’s limits (another bot or script using the same keys is the classic cause, so give SleeveFolio its own key pair).

Timeouts

Alpaca or the network. Requests that only read data (your balances, positions, and order history) are retried automatically a couple of times, all within about a second, so a momentary hiccup never reaches the screen — an error you actually see is an outage that outlasted them. Placing and canceling orders are deliberately not retried: a retry could place the same order twice, so those are attempted once and reported to you.

Check Alpaca’s status page; re-paste both halves into the matching environment card in Settings → Broker and hit Test + save once it’s green: saving re-runs the connection test against Alpaca. The automatic check that runs every 15 minutes re-compares SleeveFolio’s records with Alpaca’s and catches them up once the outage ends.

After fixing

Re-save the keys (Test + save), then check the Reconciliation page. A drift item (a recorded mismatch between SleeveFolio’s records and the broker’s) from the gap is normal; investigate and acknowledge it once you’ve confirmed the records match again.