# API changes: October 18, 2026

[← Changelog](/docs/changelog) · [October 2026](/docs/changelog/2026-10)

### FIX-1018-1: a late wake cancellation does not stop a ready server

**Before**

[POST /v1/infra/servers/:id/wake](/docs/infra/lifecycle/wake) with `wait=true` could report an error and stop an already ready server if its state changed at the end of the wait. The same affected the automatic wake of a sleeping server before [POST /v1/infra/servers/:id/deploy](/docs/infra/deploy/deploy), [POST /v1/infra/servers/:id/exec](/docs/infra/deploy/exec), [POST /v1/infra/servers/:id/upload](/docs/infra/deploy/upload) and [GET /v1/infra/servers/:id/logs](/docs/infra/deploy/logs).

**After**

If cancellation no longer applies to the server's current state, it does not stop the server. A ready server in the `running` state with a connected tunnel and no wake block returns success. If readiness cannot be confirmed, the response is `502 WAKE_FAILED`. If startup has already been refused for lack of capacity, `503 WAKE_NO_CAPACITY` is preserved. An ordinary timeout that returns the server to `sleeping` still gives `503 WAKE_TIMEOUT`.

**Impact on integrators**

The response shape and request parameters are unchanged. After `WAKE_FAILED`, check the current state through [GET /v1/infra/servers/:id](/docs/infra/servers/get) before taking the next action.

### FIX-1018-2: Creating departments in a batch

**Before:** `POST /v1/humanresources/nodes/batch` with `action: "create"` returned `ERROR_METHOD_NOT_FOUND` for every item, although individual creation worked.

**After:** creation uses REST 3.0 with a separate result and node ID for each successful item. The response remains HTTP 200; partial failure preserves the other items' results. Each item requires a separate Bitrix24 call; creation within a list is sequential to avoid conflicts on the shared structure. An uncertain write carries `details.outcome: "unknown"` and `details.retryable: false`: check the Bitrix24 account before trying again. An incomplete response without a node ID receives `WRITE_OUTCOME_UNKNOWN`. Batch update is outside this correction.