Drıftstack docs
Docs

‘The service is temporarily unavailable’ (500/502/503) and the status page

You might be here because: “The service is temporarily unavailable. Try again shortly.” · “is driftstack down” · “500 internal server error” · “502 bad gateway” · “503 service unavailable” · “An unexpected error occurred” · “everything stopped working at once” · “where is the status page” · “Driver error”

Short answer

“The service is temporarily unavailable. Try again shortly.” is what the desktop app and the dashboard say for any 500-range answer from the server. Wait a minute and retry. If it keeps happening, look at the status page, which shows overall health, each component, and open incidents.

Check first

  • Is it everything, or one action? One action failing while the rest works is more likely a switched-off feature (503) or a device problem (502) than an outage.
  • The status page shows operational, degraded (a health check failed or timed out) or major_outage; you can subscribe there to incident emails, with a confirmation link first.
  • Your own connection: “Request timed out” and “Check your connection and try again” are about your network, not the service.

Likely causes and fixes

  1. 500 — an unexpected error on the server. The API answers internal with “An unexpected error occurred.” It is the one 5xx the SDKs retry automatically, except on an AI message (message()), which they never send again by themselves. Retry; if it repeats on the same action, send the details below.
  2. 502 — the browser did not do what was asked. A driver-error names a failure in the browser running your session, not a service outage; retrying the same call will not help. Relaunch the profile.
  3. 503 — a feature switched off, not implemented for this device, no machine free just then, or the AI temporarily unavailable. driver-not-integrated is a permanent answer for this device. feature-unavailable usually means a feature is switched off on this deployment, and retrying does not change that. One case is different: on a session create, feature-unavailable with code: "dispatch_no_live_node" means no machine could take a new session just then. It carries retryable: true and a Retry-After header and nothing was created, so send the same request again after that wait. Another: on an AI message, code: "ai_provider_unavailable" means Driftstack’s included AI is temporarily unavailable and nothing ran. Wait the retry_after_seconds (also the Retry-After header, unless the turn was streamed), then send the message again with a new idempotency key if you sent one. See the feature-unavailable page.
  4. A real outage. The status page lists it as an incident with updates from investigating to resolved; wait for “resolved” before retrying a long task. The desktop app’s pill turns amber or red meanwhile — see the offline page.
  5. Dashboard says “Your sign-in has expired” instead. That is the dashboard’s wording for a 401 or 403, not a 5xx; sign in again.

If it still fails

Press Talk to a person in this chat (or open Help) and include: what you were doing, the exact sentence, the time with your time zone, your account email, and for the API the response type, detail and status. During an announced incident a reply may wait until it is resolved; nothing about your account changes because of a 5xx.