Please refer to the README for setup instructions.

Core: 5.12.2
Configuration: 4.6.0
HTTP API: 9.10.2
DreamMaker API: 6.4.4 (Interop: 5.6.0)
Web Control Panel: 4.20.0
Host Watchdog: 1.2.2

Patch 2

Fixed initial chat channel update not getting sent to the DMAPI after a TGS restart. (#1493 @Cyberboss)
Removed the possibility for chat bot connections to slow down watchdog processing upon starting or restarting TGS. (#1493 @Cyberboss)
Fixed chat bot disconnection errors being able to generate 500 responses. (#1493 @Cyberboss)
Fixed a non-fatal error registering as a 500 error when updating or restarting. (#1493 @Cyberboss)
Fixed race conditions related to updating deployment chat messages. (#1493 @Cyberboss)
Fixed a memory leak when chat bots are deleted or disabled via the API. (#1493 @Cyberboss)
Fixed a race condition relating to sending the initial watchdog chat message on instance auto-start or reattach. (#1493 @Cyberboss)

Patch 1

Upgraded git backend and removed libssl dependency on Linux. Please report any git related issues. (#1485 @Cyberboss)
Fixed an erroneous debug log when upgrading TGS. (#1485 @Cyberboss)
The server configured GitHubAccessToken will now be used to authenticate downloads of the server update package if it is present. (#1485 @Cyberboss)
Accounted for GitHub server time drift when generating the Retry-After header in HTTP 429 responses. (#1475 @Cyberboss)
Fixed an exception that can occur when performing repository jobs while using a Postgres database. (#1482 @Cyberboss)

Fixed exploit allowing for the reading of chat bot connection strings. (#1487 @MrStonedOne)

The new config version is 4.6.0. (#1461 @Cyberboss)
Added General:DeploymentDirectoryCopyTasksPerCore to allow operators to manually adjust the rate at which asynchronous file copies are performed on instances. Too few can significantly increase deployment times, too many can make TGS unresponsive and slowdown other I/O operations on the machine. Leaving this unset disables the throttle entirely. Note that this is the default for upgraded TGS installations. Server operators should highly consider setting the throttle. The default value for new installations is 100. (#1462 @Cyberboss)
Added Swarm:UpdateRequiredNodeCount option. This is the number of additional nodes a swarm controller expects to be present for an update. (#1461 @Cyberboss)

Fixed the web server not starting until the instance manager was initialized. (#1463 @Cyberboss)
Fixed a race condition that caused BYOND installation jobs to be executed in the context of the HTTP request that started them, resulting in long response times. (#1463 @Cyberboss)
Fixed modifying the system user's BYOND configuration having a race condition across instances. (#1463 @Cyberboss)
Reduced surface area for potential failure in distributed swarm server updates. (#1461 @Cyberboss)
Fixed potential deadlock if a swarm controller abandons a node that is waiting to commit an update. (#1461 @Cyberboss)

Added the ability to delete non-active BYOND installations with DELETE /Byond and the new ByondVersionDeleteRequest model. This endpoint returns an HTTP 202 job for the deletion process, HTTP 409 if the version is the active version, or HTTP 410 if the version is not installed. (#1463 @Cyberboss)
Added ErrorCode 36 for if an attempt is made to delete the active BYOND version. (#1463 @Cyberboss)
Added error code 28 for if an update can't start due to a server swarm not being fully populated (responds with HTTP 424) . (#1461 @Cyberboss)