Errors, timeouts and recovery

Check both transport errors (DNS, TLS, timeout, gRPC status) and application failures (ok=false). An empty list is meaningful only when the operation succeeded.

Symptom	Check
Cannot connect	Endpoint, TLS trust, HTTP/2, outbound firewall
Unauthorized REST/log stream	ApiKey header, real owner key and record ownership
Start refused	Broker server, login/password, permissions and ownership
Copier exists but no trade	Pause reason, event, mapping, filters, hours and margin
Wrong volume	Sizing, currency, contract size and lot constraints
No log events	Copier ID, environment, handshake and actual activity
Resume refused	Billing or another pause reason
Duplicate copying	Repeated Start after a timeout; reconcile List first

Retry deliberately

Retry transient reads with bounded exponential backoff and jitter. Do not retry invalid credentials, access denial or broker validation without fixing the cause. Reconcile state after an ambiguous Start, Pause or Remove failure before repeating a mutation.

Set a workflow deadline as well as per-call timeouts. Cancel work when its caller stops. Do not loop indefinitely against rejected broker credentials.

Report a useful issue

Include SDK commit/version, runtime version, endpoint hostname, operation, UTC timestamp, sanitized error and a minimal demo reproduction. Redact keys, passwords, raw request bodies and credential-bearing URLs.

ok=true means the management operation succeeded. It does not certify that future trades will be accepted. Verify actual positions and logs.