Skip to content

v0.2.0

Choose a tag to compare

@moritzrinow moritzrinow released this 18 Oct 19:53
· 4 commits to main since this release

Kobblestone now has a website! Feel free to check it out: https://kobblestone.io

Upgrade Notes

  • Let all running Backups and Restores finish before upgrading. Jobs created by previous versions are not considered when serializing backups and restores.
  • Apply the complete manifest of this release. The operator requires additional RBAC permissions (listing Backups, Restores and Jobs, reading PersistentVolumeClaims, managing Leases, and patching Secrets, which it grants to the auth Jobs of Accounts).
  • The operator now uses leader election via a Lease in its own namespace. After a restart or rollout, a new operator Pod waits until the lease of the previous Pod has been released or has expired (at most 15 seconds) before it starts reconciling.
  • A Server is now kept stopped while a Restore or an offline Backup of it is running, even if spec.state is Running or the server is woken up from hibernation. It is started as soon as the operation has finished.
  • The auto-sleep webhook now requests hibernation by setting the annotation kobblestone.io/hibernate: "true" on the Server instead of writing its status. The operator records the request in the Hibernating condition and removes the annotation, as it already did for kobblestone.io/wake-up.
  • status.address of Servers now contains the IP address of the server's Service instead of its cluster-internal DNS name and port (<name>.<namespace>.svc.cluster.local:<port>). It is shown in the new Address column. The Host column (the hostname of spec.route) was removed from Servers and Networks.
  • BackupRepos of type PersistentVolumeClaim now create their claim with the name from spec.persistentVolumeClaim.name. Previously, the claim was named after the BackupRepo. If those names differed, a new claim is created on upgrade. The old claim (and the snapshots in it) is left untouched and can be migrated or deleted manually.
  • Bots are now based on Minecraft-Console-Client (MCC) instead of Mineflayer. The BotBehavior resource was removed; delete all BotBehaviors and the CRD (kubectl delete crd botbehaviors.kobblestone.io). spec.behaviors, spec.viewer, spec.service and spec.auth.enableChatSigning of Bots were removed: rewrite behaviors as MCC configuration (spec.config) and scripts (spec.scripts), and use spec.auth.signChat. The <bot> Service and <bot>-viewer HTTPRoute of existing Bots are no longer managed and can be deleted.
  • Accounts have to be authorized again once. Their session Secret is replaced by one of type kobblestone.io/minecraft-session in a new, client independent format (see docs/design/0001-accounts-and-bots-on-mcc.md); the refresh token moved to the new <account>-credentials Secret. Workloads that read the old Secret keys (prismarine-auth caches) have to be adapted.

Changes by Kind

Feature

  • New Bots: a Bot is a single Minecraft-Console-Client instance connected to one Server, Network or host. It is programmed with MCC's TOML configuration (spec.config.inline and keys of ConfigMaps/Secrets in spec.config.from) and MCC command or C# scripts (spec.scripts, inline or from ConfigMaps/Secrets, optionally run on every login). Additional volumes (e.g. PersistentVolumeClaims with data files) can be mounted via spec.pod.volumes/spec.pod.volumeMounts.
  • Bots report their state in status.phase (Pending, Inactive, Connected, Disconnected), status.message (e.g. the kick reason), the resolved status.address, status.username, status.restarts and a Connected condition. Disconnected bots are restarted with backoff.
  • Bots referencing an Account stay scaled down until the Account is authorized and restart with the new session whenever the Account refreshed it.
  • New Kobblestone.AccountAuth (image ghcr.io/kobblestoneio/account-auth), replacing the Node.js implementation: performs the Microsoft device code flow and session refreshes and writes only the Account's Secrets. The Account status is derived by the operator from these Secrets and the auth Job; the Job no longer needs permissions on Accounts. It is configured with command line arguments or AUTH_-prefixed environment variables (see --help) and can also be run locally against a kubeconfig.
  • Account sessions are stored in a standard format (Secret type kobblestone.io/minecraft-session, library Kobblestone.Sessions) with the username, UUID, access token, expiry and optional chat signing key pair, so they can be mounted by any workload, not only by Bots.
  • Failed Account authorizations (e.g. an expired device code) are retried after 10 minutes, with the reason (e.g. no Minecraft profile, Xbox account restrictions) in status.message. Finished auth Jobs are kept for one hour.
  • New Kobblestone.BotRunner (image ghcr.io/kobblestoneio/bot-runner), replacing the Node.js bot runner: merges the configuration layers, hands the session over to MCC and runs it. Like Kobblestone.AccountAuth, it is configured with command line arguments or BOT_-prefixed environment variables (see --help).
  • Backups and Restores of the same Server are queued and run one after another, in the order they were created. Waiting requests do not get a Job until they can run. The running Jobs are the single source of truth, so a failed or deleted Job can no longer leave a Server locked.
  • Backups, Restores and snapshot cleanups using a BackupRepo of type PersistentVolumeClaim run one at a time per volume, unless the claim supports ReadWriteMany. This prevents Multi-Attach errors when multiple operations would run on different nodes.
  • Online backups of running servers are scheduled on the node of the Server Pod. Offline backups of stopped servers can run on any node.
  • The Server conditions BackupInProgress and RestoreInProgress now reflect the running Jobs and correct themselves on every reconcile.
  • Reasons why a resource cannot be reconciled yet (e.g. a missing BackupRepo, a Server that is not stopped or busy, an unknown Router) are now reported as Kubernetes events and, for Backups and Restores, in status.message. The resource is retried periodically instead of failing the reconciliation.
  • Backups and Restores are marked as Failed if their Job is deleted before it finished.
  • Leader election is enabled, so the operator can safely run with multiple replicas.
  • Every resource's status is now written only by its own controller. Controllers watching child resources (Jobs, StatefulSets, Services, Commands) queue the owning resource for reconciliation instead of updating its status, which prevents conflicting and lost status updates.
  • Routers, Servers and Networks report the IP address of their Service in status.address (shown in the Address column), preferring external addresses: the IP assigned by a load balancer, then the first of spec.externalIPs, then the cluster IP. The address is updated as soon as the Service changes, e.g. when a load balancer assigned an IP.
  • All fields of all resources are now documented in the CRDs (shown by kubectl explain and in the CRD reference docs).
  • Backups and Restores have a new status field startedAt, the time their Job was created.

Bug or Regression

  • Fix Routes being deleted right after their creation when a Server or Network without spec.route has the same name. Servers and Networks now only update or delete the Route they created for spec.route.
  • Fix two pending Backups of the same Server blocking each other forever.
  • Fix Restores not waiting for running Backups or other Restores of the same Server.
  • Fix online backups failing with Multi-Attach errors when the backup Pod was scheduled on a different node than the Server.
  • Fix backup Pods being blocked from accessing RCON by the NetworkPolicy of the Server.
  • Fix reconciliation errors for Backups and Restores whose Job already existed (e.g. after an operator restart).
  • Fix a stale BackupInProgress condition blocking the deletion of a Server forever.
  • Fix Routes not being retried when their Router did not exist yet.
  • Fix Commands not being retried when the RCON password secret did not exist yet.
  • Fix BackupRepos of type PersistentVolumeClaim creating their claim with the name of the BackupRepo instead of spec.persistentVolumeClaim.name.
  • Fix Restores ignoring spec.job.image.
  • Fix Networks ignoring spec.proxy.offline. If set to true, the proxy now runs with online-mode = false.
  • Fix the documentation of Server spec.lifecycle.rolloutPolicy, which listed OnRestart instead of the actual value Manual.
  • Fix Routers reporting their cluster IP when the first load balancer ingress entry had a hostname instead of an IP.
  • Fix the auto-sleep webhook never hibernating a Server (namespace and name were swapped when looking up the server).
  • Fix Servers never hibernating again after being woken up, because the kobblestone.io/wake-up annotation was never removed.
  • Fix Backups with retainPolicy: RequireDelete never retrying a failed cleanup. Deleting the failed cleanup Job now starts a new one.

Dependency

  • Bump KubeOps to 13.3.3.
  • Bump KubernetesCRDModelGen.Models.gateway.networking.k8s.io to 1.57.0.
  • Bump itzg/mc-backup to 2026.10.0 for Backups.
  • Pin itzg/mc-proxy to java25-2026.10.0 for Networks.
  • Bump itzg/mc-router to 1.48.0 for Routers and Server sidecars. With autoSleep.mode: Hibernate, only player logins reset the scale-down timer now, so server list pings no longer keep a server awake.
  • Bump itzg/minecraft-server to 2026.10.0 for Servers.
  • Bump amazon/aws-cli to 2.37.13 for Backups, Restores and snapshot cleanups using S3 repos. Checksums are only sent when required, to stay compatible with S3-compatible providers.
  • Bump ubuntu to 26.04 (from the end-of-life 25.04) for Backups, Restores and snapshot cleanups.
  • Replace the Node.js images account-auth and bot-runner (Mineflayer, prismarine-auth) by .NET images in version 0.2.0. The bot image bundles an unmodified Minecraft-Console-Client release (20260929-519, CDDL-1.0).