Stormancer.Server.Node 9.2.1.15-pre

• net8.0

  • Autofac (>= 8.0.0)
  • Autofac.Extensions.DependencyInjection (>= 9.0.0)
  • CommunityToolkit.HighPerformance (>= 8.2.2)
  • MessagePack (>= 2.5.140)
  • Microsoft.AspNetCore.Mvc.NewtonsoftJson (>= 6.0.10)
  • Microsoft.IO.RecyclableMemoryStream (>= 3.0.0)
  • Mono.Posix.NETStandard (>= 5.20.1-preview)
  • Newtonsoft.Json (>= 13.0.3)
  • NuGet.Protocol (>= 6.9.1)
  • Portable.BouncyCastle (>= 1.8.8)
  • RocksDB (>= 8.10.0.45817)
  • SmartFormat.NET (>= 3.3.2)
  • Spectre.Console (>= 0.44.0)
  • Stormancer.Abstractions.Core (>= 2.4.1.6-pre)
  • Stormancer.Abstractions.Host (>= 13.0.0-pre)
  • Stormancer.Abstractions.Logging (>= 1.0.1.1-pre)
  • Stormancer.CommandLine (>= 1.2.2)
  • Stormancer.Filters (>= 0.2.0.2-pre)
  • Stormancer.Packaging (>= 3.0.1.1)
  • Stormancer.Raknet.Net (>= 1.1.3-pre)
  • Stormancer.Server.Common (>= 3.0.1-pre)
  • Stormancer.Tcp (>= 6.0.0.23-pre)
  • Swashbuckle.AspNetCore (>= 6.5.0)
  • System.Diagnostics.DiagnosticSource (>= 8.0.0)
  • System.IdentityModel.Tokens.Jwt (>= 7.4.0)
  • System.Interactive.Async (>= 6.0.1)
  • System.IO.Abstractions (>= 20.0.28)
  • System.Reactive (>= 6.0.0)
  • System.Runtime (>= 4.3.1)

Added
*****
- Added Development mode to applications, enabling running them without deployment in this mode. (an app process must be launched externally in this case)
- Added an API to force a shard to become primary to fix a cluster where a node hosting a primary was definitely killed without the system being able to bring its shards down.
- Cluster settings enable to set configurations settings that are synchronized across the cluster.
- Added levels to the health check API.
- Fix issue that allowed Admin APIs to be available on public endpoints.
- Add configuration option `webApis.SslProtocols` to control which Ssl protocols are supported by the web API.
- Added Leader metrics to monitor cluster state, leadership and node count in the cluster.
- Added `allocation.scenes.useShardLocation` switch in the cluster settings to force the cluster to allocate scenes in the same node as the corresponding shard. Defaults to false.
- Added live informations about where the scene is allocated to the get scene admin Web API.
- Added a web api to set a single configuration setting.

Changed
*******
- Refactored the cluster WebApi system to support https on public and admin APIs and JWT bearer token authentication.
- Reworked the node configuration for web apis. WARNING: Older web api configurations are obsolete, and are not supported anymore.
- Removed the Git based deployment APIs. They have been deperecated for a long time.
- Do not emit a connection failure log each second when failing to contact a node in the endpoints discovery list.


Fixed
*****
- When failing to start the N2N endpoint, the autojoin logic would be started concurrently again and again without stopping the previous one, leading to a task leak.
- The Elector shouldn't start if the node fails to bind to the N2N endpoint.
- Blob storage: When a block blob is missing in a block, return an error. The system instead entered a softlock.
- Fixed a race condition that could lead a failed Host entry to be kept in the hostmanager.
- When nodes are offline in the cluster, they are now ignored when doing a cluster wide system request to prevent spurious unavailability errors, for instance on deploy.
- Fixed an issue that could prevent scene disconnect events to be reported to peers correctly in a multinode environment if the scene and the peer lived on different nodes.
- Fixed an issue that could prevent server triggered scene disconnects to trigger a disconnected event on the scene serverside.
- Fixed a memory leak in the Application records caching system.