v0.43.2

v0.43.2 - 22 Apr 2024

• vmagent: fixes bug with ServiceAccount not found with ingestOnlyMode.
• vmagent: fixes unknown long flag '--rules-dir' for prometheus-config-reloader.

v0.43.1

Full Changelog: v0.43.0...v0.43.1

v0.43.0

v0.43. - 18 Apr 2024

Update note: vmcluster: remove fields VMClusterSpec.VMInsert.Name, VMClusterSpec.VMStorage.Name, VMClusterSpec.VMSelect.Name, they're marked as deprecated since v0.21.0. See this pull request.
Update note: PodSecurityPolicy supports was deleted. Operator no long creates PSP related objects since it's no longer supported by Kubernetes actual versions. See this doc for details.
Update note: PodDisruptionBudget at betav1 API is no longer supported. Operator uses v1 stable version. See this doc for details.
Update note: Alertmanager versions below v0.22.0 are no longer supported. Version must upgraded - manually for resources or use default version bundled with operator config.

• operator: properly reconcile ServiceAccount specified for CRDs. Previously operator didn't perform a check for actual owner of ServiceAccount. Now it creates and updates ServiceAccount only if this field is omitted at CRD definition. It fixes possible ownership race conditions.
• Update VictoriaMetrics image tags to v1.100.1.
• operator: reduce number of watched resources owned by CRDs. Operator no longer watches for Service, Secret, Configmap changes owned by CRD object. It must reduce logging output, CPU and memory usage for operator.
• operator: exposes config-reloader-http port with 8435 number for the customer config-reloader containers. Operator may use own config-reloader implementation for VMAuth, VMAlertmanager and VMAgent.
• operator: adds new field configReloaderExtraArgs for VMAgent, VMAlert, VMAuth and VMAlertmanager CRDs. It allows to configure config-reloader container.
• config-reloader: adds error metrics to the config-reloader container - configreloader_last_reload_successful, configreloader_last_reload_errors_total, configreloader_config_last_reload_total, configreloader_k8s_watch_errors_total, configreloader_secret_content_update_errors_total, configreloader_last_reload_success_timestamp_seconds. See this issue for details.
• operator: Changes error handling for reconcile. Operator sends Events into kubernetes API, if any error happened during object reconcile. See this issue for details.
• operator: updates base Docker image and prometheus_client to versions with with CVE fixes
• operator: adds reconcile retries on conflicts. See this issue for details.
• operator: allows adjust Service generated by operator with useAsDefault option set to true for serviceSpec field. See this issue for details.
• vmagent: allows to modify serviceName field for vmagent at statefulMode with custom service. See this issue for details. Thanks @yilmazo
• vmagent: change service for statefulMode to the headless instead of clusterIP. See this issue for details.
• vmservicescrape&vmpodscrape: add attach_metadata option under VMServiceScrapeSpec&VMPodScrapeSpec, the same way like prometheus serviceMonitor&podMonitor do. See this issue for details.
• vmagent: allows multi-line regex at relabelConfig. See this docs and this issue for details.
• vmalertmanagerconfig: fix struct field tags under Sigv4Config.
• vmalertmanagerconfig: adds own config-reloader container. It must improve speed of config updates. See this issue for details.
• vmalertmanager: bump default alertmanager version to v0.27.0, which supports new receivers like msteams_configs.
• vmalertmanager: supports alertmanager version v0.22.0 or higher. Previous versions are no longer supported and must be upgraded before using new operator release.
• vmscrapeconfig: add crd VMScrapeConfig, which can define a scrape config using any of the service discovery options supported in victoriametrics.
• vmuser: adds targetRefBasicAuth field targetRef, which allow to configure basic authorization for target_url. See this issue for details. Thanks @mohammadkhavari
• vmprobe: add field proxy_url, see this issue for details.
• scrape CRDs: add field series_limit, which can be used to limit the number of unique time series a single scrape target can expose.
• scrape CRDs: fix scrape_config filed disable_keep_alive, before it's misconfigured as disable_keepalive and won't work.
• scrape CRDs: deprecated option relabel_debug and metric_relabel_debug, they were deprecated since v1.85.0.

v0.42.4

Full Changelog: v0.42.3...v0.42.4

v0.42.3

v0.43. - 12 Mar 2024

• vmalert: do not add notifiers.* flags in case notifier.blackhole is provided via spec.extraArgs. See this issue for details.
• operator: properly build liveness probe scheme with enabled tls. Previously it has hard-coded HTTP scheme. See this issue for details.
• operator: do not perform a PVC size check on StatefulSet with 0 replicas. It allows to creates CRDs with 0 replicas for later conditional resizing.
• vmalertmanager: properly print columns at CRD replicaCount and version status fields.

v0.42.2

v0.42. - 6 Mar 2024

• operator: fixes alertmanager args typo.
• prometheus-converter: adds new flag controller.prometheusCRD.resyncPeriod which allows to configure resync period of prometheus CRD objects. See this issue for details.

v0.42.1

v0.42. - 5 Mar 2024

• operator: properly watch for prometheus CRD objects. See this issue for details.

v0.42.0

v0.42.0 - 4 Mar 2024

• operator: adds more context to the log messages. It must greatly improve debugging process and log quality.
• Update VictoriaMetrics image tags to v1.99.0.
• operator: allow multiple comma separated values for WATCH_NAMESPACE param. It adds multiple watch namespace mode without cluster-wide permission. See this issue for details. Need namspace RBAC permissions located at config/examples/operator_rbac_for_single_namespace.yaml
• operator: updates runtime dependencies (controller-runtime, controller-gen). See this issue for details.
• operator: updates runtime dependencies (controller-runtime, controller-gen). See this issue for details.
• operator: adds new status.updateStatus field to the all objects with pods. It helps to track rollout updates properly.
• operator: adds annotation operator.victoriametrics/last-applied-spec to all objects with pods. It helps to track changes and implements proper resource deletion later as part of issue.
• vmagent: adds flush_on_shutdown to the streamAggrConfig. See this issue for details.
• vmagent: adds spec.ingestOnlyMode experimental field. It switches vmagent into special mode without scrape configuration and config-reloaders. Currently it also disables tls and auth options for remoteWrites, it must be addressed at the next releaes.
• vmalertmanager: use blackhole as default router if not configuration provided instead of dummy webhook. 9ee567f
• vmalertmanager: properly assign path for templates, if it's configured at config file and defined via spec.templates. 1128fa9
• vmauth: adds new field spec.configSecret, which allows to use vmauth with external configuration stored at secret under config.yaml key. Configuration changes can be tracked with extraArgs: configCheckInterval: 10s or manually defined config-reloader container.
• vmstorage: properly disable pvc resizing with annotation operator.victoriametrics.com/pvc-allow-volume-expansion. Previously it was checked per pvc, now it's checked at statefulset storage spec. It also, allows to add pvc autoscaler. Related issues #821, #867.
• [apis]: add missing static config relabeling config and test case by @incubator4 in #884

New Contributors

• @incubator4 made their first contribution in #884

Full Changelog: v0.41.2...v0.42.0

v0.41.2

v0.41.2 - 21 Feb 2024

• Remove deprecated autoscaling/v2beta1 HPA objects, previously operator still use it for k8s 1.25. See this issue for details.
• Update VictoriaMetrics image tags to v1.98.0.

v0.41.1

v0.41.1

• update VictoriaMetrics image tags to v1.97.1.