You can pull Kura 0.10.5 as ghcr.io/tuist/kura:0.10.5.
Reliability
- Remote cache replication pods now finish bootstrap for accounts whose cache exceeds the temporary-directory budget. Previously, concurrent peers staged artifacts into the same shared tmp directory without reserving space, so the directory could exhaust its budget and leave pods stuck
NotReadywith repeated 503 readiness checks. Kura now holds a per-artifact byte reservation, so peak staging stays within the budget and bootstrap converges.
Operations
- Kubernetes workload clusters now prefer kubelet
InternalIPaddresses, restoringkubectl logsandexecon nodes that have no ExternalIP or resolvable hostname. This makes it easier to diagnose stuck remote cache pods.