Large sites no longer freeze for about a second when the game moves a workload, you change a cable, or you rename a device. A Kubernetes cluster short of a node now gets it once a server has room.
Fixed
Performance
- Workload moves no longer freeze large sites. Draining a server now runs without the game stalling every few seconds.
- Cable changes no longer freeze large sites. Laying, pulling or re-seating a cable still moves traffic onto the new routes straight away.
- Renaming a device no longer freezes the game. The new name still shows at once on the Network page.
Customers
- A Kubernetes cluster short of a node now gets it. The game places the missing node on any server with room, instead of waiting for room to rebuild the whole cluster.
- The capacity note names a real server. It states what the closest server is short of, instead of reading 0 beside servers with room.
- Diagnose says where a customer's copies are stacked. The Copies kept apart row names the machine, rack or zone and how many copies stop if it fails, for every service.
Security and networking
- A SIEM plugged into a powered-off switch no longer counts as working. It stops meeting a customer's audit-logging requirement until the switch is back on.
- Two services of one customer on one bonded server keep their own traffic split. The server's link loads now match the traffic each service sends.
Notes
- Saves load both ways. No save fields changed.