2.0.0 (2026-01-28)

Bug Fixes

• add missing build information (ddeab5a)
• allow auto creation of registry monitors for registries with a port (b2bd000)
• alternatives list building (5f64279)
• chart: default repositories in values is quay.io (b35e396)
• config: set default routing check timeout to 2s in the code (8782900)
• consider ratelimit-remaining header from docker hub while checking image availablity (e6d4ca2)
• controller: CISM and ISM should exclude their own mirrors to avoid mirror loops (478fce2)
• controller: ignore digest-based images (fc2f07d)
• controller: prevent reconciliation loop (34a5c32)
• controllers: add new mirrors from (Cluster)ImageSetMirror spec to its status (31b3eab)
• controllers: don't log about filtering when there is 0 image to filter (a6aa7dc)
• controller: set mirroring exponential backoff baseDelay to 1s (80aa6cc)
• copy credentialprovider from k8s repo and upgrade dependencies (715690f)
• create RegistryMonitors for alternative routing/mirroring registries too (3bb7ffb)
• deduplicate filtered out images in mirror loop protection logs (2817eda)
• delete images by digests (8709b93)
• don't check for image availability if there is no alternative configured (347277b)
• don't create the same RegistryMonitor twice (51dff02)
• don't log "successfully mirrored image" on mirroring failure (3d22a6c)
• don't mutate twice the same container nor mutate a pod on update (8327d81)
• don't use generatedNames in RegistryMonitors as it still can produce duplicates specs (9a2fffa)
• expires images from (Cluster)ImageSetMirrors status that doesn't match anymore (569d2c2)
• GetKeychains always returns at least 1 keychain (74f6e05)
• helm: add missing Age printcolumn in RegistryMonitor (091a596)
• helm: add missing permissions for (Cluster)ReplicatedImageSets (9b64272)
• helm: allow manager to get, list and watch secrets (6953d32)
• helm: always create MutatingWebhookConfiguration (2378f11)
• helm: better default values for registry monitors and metrics (870852a)
• helm: lower default throttling values for RegistryMonitors and add registry.k8s.io and public.ecr.aws (36e5ece)
• helm: rbac (b575666)
• helm: remove metrics (5359cb9)
• helm: set default unusedImageTTL to 24 hours (0185ae1)
• helm: values (278ef70)
• imageFilter include .* by default only when at least one exclude pattern is defined (d87364a)
• improve logs and rename badly named images (d5e4458)
• improve monitoring logs (monitoring failed is not an error but an info) (aa34ec5)
• include registry while matching images against (Cluster)ReplicatedImageSets filters (c701d7e)
• linter errors (fb45e08)
• metrics: init registry_monitor_tasks_total and registry_monitor_images to 0 for all registries and status (bb84c24)
• metrics: label registry_monitor_registries metrics with registry spec, not name as they may be hashed values (bb87f05)
• metrics: remove registry_monitor_registries metrics as it not used anymore (1794769)
• panic in pod webhook when an alternative image doesn't have a secret associated (58f8d4d)
• panics in registry requests when no timeout has been configured (faf07f9)
• platform amd64 is not recognized (should be linux/amd64) (50aa6e7)
• prevent ignoring throttling when maxPerInterval > len(images) (3ddee0b)
• prevent mirror loops (8ae21ea)
• prevent reconciliation loop when health checking a registry (efb8068)
• QuotaExceeded are no longer reported as Unreachable (f0178f9)
• rbac: allow to update CRIS (490868f)
• recreate ImageMonitors on deletion when the required Image and RegistryMonitor exist (76b5e6b)
• remove references to already removed field RegistryMonitor.status.registryStatus (c42d2f3)
• remove RegistryMonitor CRD (0349822)
• requeue (Cluster)ImageSetMirrors after expiration duration (0f9e026)
• set a default value for registymonitor.interval (10m) (a100b7f)
• set status Available instead of MissingSecret for images without pods that don't need authentication (b29cc2d)
• skip deletion of images not mirrored yet (6153b5b)
• skip monitoring images with no pods to prevent the random picker to crash (64f962e)
• timeout registry requests per keychain (bffcaf2)
• typo in role for ReplicatedImageSet ressources (034f518)
• use HEAD instead of GET when monitoring an image in to by-pass docker hub quotas (23c2253)
• use http GET instead of HEAD as ghcr.io doesn't support HEAD /v2/ (58c6020)
• use one monitor pool per registry (31b91f2)
• use pool.Submit instead of pool.Go in order to prevent silent panics (01c7696)
• watch Pods to reconcile (Cluster)ImageSetMirrors (3daa053)
• webhook: authenticate active monitoring requests (d4d45e4)
• webhook: ignore digest-based images (0aef45c)
• webhook: ignore invalid images (79c7010)
• webhook: match prefix on normalized image (efca35c)
• webhook: resolve race condition in alternatives check result ordering (c7c4160)

chore

• wipe repository for v2 (ffccda2)

Features

• add a status "Unreachable" in image upstream status and record last error in the image status (8031cc0)
• add image upstream status (4d6a633)
• add LastError field in image mirroring status (b08691f)
• add QuotaExceeded image status (394a92f)
• add registryMonitor.status.registryStatus to keep track of registries health (531df0f)
• add Scheduled and UnavailableSecret Image monitoring status (db5f70b)
• add timeout to monitoring tasks (active and passive) (70e8756)
• basic image mirroring (no-auth, amd64 only) (75ced26)
• basic image monitoring without authentication support (13362fd)
• cleanup (Cluster)ImageSetMirror before deletion (8453b5f)
• cleanup secrets on CISM/CRIS deletion (405a1de)
• copy imagePullSecrets in pod's namespace when using cluster-scoped ISM and RIS (af21210)
• create ImageMirrors for Images that match a strategy from the configuration (9a1dddc)
• create images from nodes instead of pods and update their status (e574d53)
• create Images from Pods (1a28e55)
• create one ImageMonitor per matching registry based on routing strategies (c7a5392)
• delete expired images (10f42f4)
• delete unused images after a configurable amount of time (c6f6813)
• helm chart (3506acf)
• helm: add support for topologySpreadConstraints (de59664)
• helm: allow to populate the config.yaml file (1c71b82)
• helm: create default registry monitors on installs and upgrades (fd3c68a)
• helm: don't leader elect if replicas < 2 (633bd17)
• helm: merge registryMonitors.items default values with user supplied values and make it a dict instead of an array (1eb67c8)
• imageMatcher becomes a list of include and exclude regexps (0b676ee)
• inject rerouted image pull secret if not already present in the pod (bc45e1d)
• load configuration from file (7625881)
• make default http method per registry monitor configurable (HEAD/GET) (5a27928)
• mark unused images and don't monitor them (e2ef3bb)
• metrics: add detailed status in registry_monitor_tasks_total metric (a73527d)
• metrics: add image_last_monitor_age_minutes histogram metric (a0f4e21)
• metrics: add registry_monitor_images metric (6ccb41b)
• metrics: add registry_monitor_registries metric (2b084f7)
• metrics: add registry_monitor_tasks_total metric (b86afd8)
• metrics: add unused label on images and tasks (e96989a)
• metrics: setup prometheus exporter and export manager metrics (2de00fd)
• mirror images according to ClusterImageSetMirrors (3a371ae)
• mirror images according to ImageSetMirrors (3d1737f)
• mirroring with auth (aa59c20)
• optionnal active check when rerouting images (2415181)
• recreate images when deleted while being used (575b1c1)
• registry monitors auto-discovery (create registry monitors with default values for registries used in the cluster) (f62f32f)
• reroute images based on (Cluster)ImageMirrorSets (d2553f7)
• reroute images based on (Cluster)ReplicatedImageSets (bb1c312)
• reroute images based on their availability (b35babd)
• set unusedSince status in (Cluster)ImageSetMirrors when an image stopped to be used in the cluster (2586a9b)
• use secrets configured in ReplicatedImageSet & ImageMirrorSets to check images availability (70e06f7)
• use secrets from pod to authenticate image scans (aac91fb)
• webhook: cache choosen alternatives as well as manifests availability (89031a8)
• webhook: cache manifests HEAD results (ef7a5db)
• webhook: run checks in parallel and lower timeout to 1s (937c011)

BREAKING CHANGES

• rebooting the project!