Note: These changes apply to the runtime of all SpatialOS 11+ SDK versions.
Next time you deploy locally or remotely, you’ll automatically receive the latest version, so you don’t need to do anything to upgrade.
- Rewrote error messages returned by failing entity command response ops. The error messages for commands failing with “timeout”, “not found” and “authority lost” status codes now mention whether the command request was ever forwarded to the target worker and began executing, or whether the command request failed before it made it to the worker. We took the chance to revamp our commands documentation alongside this change: https://docs.improbable.io/reference/14.0/shared/design/commands
- Improved how quickly the runtime detects workers disconnecting from the deployment. For users that disconnect and reconnect their server workers using the ‘manual_worker_connection_only’ option, this should significantly reduce instances of the load balancer giving no work to the newly reconnected worker because the old worker was still thought to be connected.
- Fixed a bug where connecting both KCP and modular UDP connections to the same deployment could cause a crash.
When dynamically adding and removing a component again in quick succession, the authoritative worker can receive
RemoveComponentops for this component that it isn’t supposed to see.
Authority handover timeouts for authority loss imminent notifications are not applied on a per-worker, per-component basis as documented. Instead,
authorityHandoverTimeoutMsvalues in the bridge configuration for any worker type currently apply to all worker types. In particular, this means that setting a
defaultsection authority timeout will enable authority loss imminent notifications for all components on all worker types, unless overridden in the