Skip to main content
product · changelog

Changelog

Every release, every fix, every new integration. We ship every two weeks.

RSS Feed

AWS Databricks support

ZopNight extends Databricks support to AWS — connect your AWS Databricks workspace as a separate connection and ZopNight manages it end-to-end, from discovery and cost to savings recommendations and start/stop scheduling. ZopDay also lets admins create and edit Secret and Namespace manifests directly from the Kubernetes resource views.

  • New ZopNight: AWS Databricks support — connect your AWS Databricks workspace as a separate connection. ZopNight discovers clusters, instance pools, SQL warehouses, jobs, and model-serving endpoints with per-resource cost, recommends savings on idle instance pools and oversized clusters (priced against real pricing), and can start and stop clusters, instance pools, and SQL warehouses
  • New ZopDay: admins can now create and edit Secret and Namespace manifests directly from the Kubernetes resource views, with safety guards on the apply path to prevent accidental edits

AWS Bedrock and GCP Vertex AI support

ZopNight now covers AWS Bedrock and GCP Vertex AI end-to-end — discovering and managing the full estate of agents, models, knowledge bases, feature stores, endpoints, and jobs, with per-resource cost and metrics. Schedule recommendations can be applied in one click, and recommendation coverage and accuracy both widen across AWS and GCP.

  • New ZopNight: full AWS Bedrock support — discovers and manages agents, knowledge bases, guardrails, foundation/custom/imported models, provisioned throughput, flows, and customization, batch-inference, and evaluation jobs, with per-resource cost and metrics
  • New ZopNight: full GCP Vertex AI support — endpoints, models, indexes, feature stores, datasets, Workbench notebooks, reasoning engines, and training, tuning, batch-prediction, and pipeline jobs, with per-resource cost and metrics
  • New ZopNight: broader recommendation coverage — license-cost advisories (AWS + GCP), SageMaker GPU-idle endpoints, a revived NAT → VPC-endpoint advisory, and more orphan-cleanup rules (idle dashboards, unused alarms, idle GCP load balancers, abandoned S3 multipart uploads)
  • Improved ZopNight: apply a Schedule recommendation in one click — no need to recreate the schedule by hand
  • Improved ZopNight: more accurate recommendations — wider metric coverage (memory, GPU, and database-capacity signals across AWS and GCP) and a steadier 30-day scoring window reduce noisy or premature recommendations

Azure Databricks savings recommendations

ZopNight extends savings recommendations across your Azure Databricks compute — clusters, instance pools, SQL warehouses, jobs, and model-serving endpoints — covering rightsizing, spot discounts, governance, idle, and orphan cleanup. Idle recommendations also get safer and more accurate, and ZopDay can now show the logs of a container's previous, terminated instance.

  • New ZopNight: savings recommendations now cover Azure Databricks — clusters, instance pools, SQL warehouses, jobs, and model-serving endpoints — across rightsizing, spot discounts, governance, idle, and orphan cleanup
  • Improved ZopNight: idle recommendations are safer and more accurate — they default to scheduling instead of a one-shot stop, never suggest terminating active resources, and savings no longer double-count across recommendation types
  • Improved ZopDay: the cluster log viewer can now show the logs of a container's previous, terminated instance — useful for debugging a crash or restart

Full AWS SageMaker coverage, plus savings on the cost trend graph

ZopNight now covers AWS SageMaker end-to-end — 11 more resource types discovered automatically, jobs costed by how long they actually ran, recommendations across endpoints and clusters, and HyperPod clusters schedulable on and off. The Cost Reports trend graph also draws daily savings as its own line.

  • New ZopNight: full AWS SageMaker coverage — 11 more types discovered automatically, jobs costed by actual runtime, and HyperPod clusters schedulable on and off (existing AWS customers grant new read permissions for full coverage)
  • New ZopNight: recommendations now cover SageMaker — idle and over-provisioned endpoints and clusters, batch jobs, managed-spot training, and security checks
  • New ZopNight: the Cost Reports trend graph now draws your daily savings as its own line alongside spend
  • Improved ZopNight: remediation failures now explain why in plain language, without leaking cloud identifiers, and show an actionable grant card with a console deep link
  • Improved ZopNight: Cost Reports → Tags paging is faster and lighter on accounts with many tags

Kubernetes recommendations, invoice-accurate cost, realistic savings

ZopNight recommendations now extend to your Kubernetes clusters — health, autoscaling, storage, networking, and security — on a new Recommendations page, each with a ready-to-run playbook. Savings figures now track your real bill, and cost totals reconcile with your cloud invoice.

  • New ZopNight: Kubernetes recommendations for your clusters — health, autoscaling, storage, networking, and security — on a new Recommendations page, each with a ready-to-run kubectl/yaml playbook
  • Improved ZopNight: savings figures now match your real bill — part-time resources are no longer over-stated, totals can't exceed your bill, and more resource types are covered
  • Improved ZopNight: cost totals now reconcile with your cloud invoice — data transfer, taxes, fees, credits, and refunds are all counted, so headline totals may exceed per-team and per-tag breakdowns by design
  • Improved ZopNight: new overrides inherit the target's schedule timezone (shown in a tooltip), and you're warned upfront if the target already has an active override
  • Improved ZopNight: GCP VM install now uses the IAP SSH permission — existing GCP customers re-authorize their cloud account to enable it

Auto-remediation expands into rightsizing and config — ~132 rules

ZopNight auto-remediation moves beyond stop and idle cleanups into a new rightsizing and config rule class, ~132 rules across AWS, Azure, and GCP, with all database-config rules guided (type-to-confirm). Architecture and Cost Reports also gain a unified tag picker that groups values under collapsible keys.

  • New ZopNight: a new rightsizing and config rule class takes auto-remediation to ~132 rules across AWS, Azure, and GCP — covering storage tiering, log retention, Lambda, RDS, and managed databases
  • New ZopNight: all database-config rules are guided (type-to-confirm)
  • Improved ZopNight: a unified tag picker on Architecture and Cost Reports → Tags groups values under collapsible keys, with search and per-value cost

VM observability for ZopDay across AWS, Azure, and GCP

Services running on ZopDay VM deployment spaces now have live metrics, logs, and events across AWS, Azure, and GCP, at parity with the cluster observability tab and with no SSH or public IP required. Quick Deploy adds repo-picker, GitHub PAT, and per-deployment hosts, and org-member search now runs server-side.

  • New ZopDay: live metrics, logs, and events for services on VM deployment spaces across AWS, Azure, and GCP, at parity with the cluster observability tab — no SSH or public IP needed
  • Improved ZopDay: Quick Deploy now works with repos picked from any connected integration, and adds GitHub PAT support and a unique public host per deployment
  • Improved ZopNight: the onboarding connect flow now uses the same steps as settings, fixing the stuck Azure Discovery Scope selector
  • Improved ZopNight & ZopDay: org-member search now runs server-side across all members by name or email

Kubernetes node and workload cost breakdown

ZopNight now breaks cluster cost down to the workload — per-namespace and per-deployment spend across AWS (EKS), GCP (GKE), and Azure (AKS), including how much you pay for but don't use. It's a breakdown of the cluster's existing VM spend (reservations, savings plans, and spot included), not new charges, so cost summaries and the cloud bill never double-count.

  • New ZopNight: per-workload Kubernetes cost breakdown across AWS (EKS), GCP (GKE), and Azure (AKS) — see which namespace and deployment drives spend, and how much is idle waste
  • New ZopNight: workload cost is a breakdown of the cluster's existing VM spend (reservations, savings plans, and spot included), not new charges, so summaries and the cloud bill never double-count

Import existing Helm apps, plus 28 one-click idle cleanups

ZopDay can now adopt the Helm apps already running on your clusters with zero redeploy, keeping the cluster read-only. ZopNight completes IAM Import with Azure, turns 28 idle recommendations into one-click cleanups, and attributes child-resource cost more accurately through tag inheritance. Azure VM deployment reaches parity with AWS and GCP, and the Architecture Canvas gains dark mode.

  • New ZopDay: import the Helm apps already running on your clusters into ZopDay's project, environment, and service model, with the cluster kept strictly read-only
  • Improved ZopNight: Azure IAM Import now goes end-to-end, joining AWS and GCP — built-in and custom roles are translated, and anything that cannot be mapped safely is flagged
  • Improved ZopNight: one-click cleanup for idle resources — 28 recommendations are now one-click (6 automatic, 22 guided), with idle managed databases getting a power-only pause
  • Improved ZopNight: more accurate cost attribution — untagged child resources inherit their parent's tags, marked with a green inherited badge
  • Improved ZopNight: the Architecture Canvas now fully supports dark mode, with smoother zoom and pan on large accounts
  • Improved ZopDay: build and deploy services onto Azure VMs, bringing Azure to parity with AWS and GCP for VM deployments
  • New Internal: a public liveness endpoint lets external uptime monitors confirm each service is up (ops-only, no customer-facing change)

Automatic IAM translation, Azure anomaly hierarchy, and 13 new cleanup rules

IAM imports for AWS and GCP now translate cloud roles automatically, flagging only the actions that genuinely need review. Azure cost-anomaly detection extends to Resource Group and Tenant levels, thirteen more orphaned and idle resources clean up automatically, and the cloud-accounts page spells out each account's permission state. ZopDay adds full AWS VM provisioning and deployment.

  • Improved ZopNight: automatic IAM role translation for AWS and GCP imports, flagging only the actions that could not be translated for review
  • New ZopNight: Azure cost-anomaly detection now runs at the Resource Group and Tenant levels, not just the Subscription level
  • New ZopNight: 13 new orphan and idle cleanup rules with auto-remediation across AWS and Azure, now including free and zero-cost orphans
  • New ZopNight: the cloud-accounts page shows each account's permission state — Probe-blocked, Write access not set up, or Read-only — with guided fixes
  • New ZopDay: provision a complete AWS VM environment and deploy git- or image-based services onto it, with custom domains, HTTPS, and private registries
  • Improved ZopNight: app-wide tooltip polish, Cost Over Time from day 1, smarter Create Override defaults, and mobile sidebar scroll locking

ZopDay deploys to VMs, alongside Kubernetes

ZopDay now deploys to VMs as a first-class target alongside Kubernetes: push code, the image builds, and the service runs behind a managed L4 load balancer with TLS ingress. VMs sit in private subnets with no open management ports, one VM can host many services, and the deployer now detects the environment variables a service needs during stack detection.

  • New ZopDay: deploy services to VMs alongside Kubernetes — push code, it builds, and runs behind a managed L4 load balancer with TLS-terminated ingress
  • New ZopDay: VMs run in private subnets with no external IP and no open management ports (control plane via Google IAP / AWS SSM / Azure Run-Command)
  • New ZopDay: one VM can host many services, custom domains attach from the UI with automatic DNS, and TLS is issued via Let's Encrypt
  • New ZopDay: a VM-pool detail page shows health, attached domains, live deployments, and provisioning history
  • Improved ZopDay: stack detection now reads the environment variables a service needs from the codebase and .env files and pre-fills the deploy wizard (Node.js, Python, Go, Java/Kotlin, Ruby, PHP)

Permission visibility, GCP IAM import, and storage lifecycle remediation

When ZopNight can't read part of your cloud, the page that needs it now says so: Resources, Schedules, Recommendations, and Cost Reports each show a banner naming the missing access and what it would unlock, and a new View Permissions screen brings every permission ZopNight checks into one place, tracked per account, per region, and per cloud. GCP IAM import goes end-to-end, one-click storage-lifecycle remediation lands across all three clouds, and ZopDay gets a live Observability tab per service.

  • New ZopNight: per-page permission banners on Resources, Schedules, Recommendations, and Cost Reports name the specific access that's missing and the resources it would unlock
  • New ZopNight: a View Permissions screen brings every permission ZopNight checks into one place, grouped by feature and tracked per account, per region, and per cloud, so a partial setup is visible instead of rolled up to one status
  • New ZopNight: GCP IAM import end-to-end (mirroring AWS): one Cloud Asset Inventory call returns the project's IAM policy, service accounts are filtered out, and Google Group members can optionally be resolved via Cloud Identity
  • New ZopNight: GCP onboarding gets a dedicated BigQuery cost-export step, also offered as a one-click fix when GCP cost data is missing after onboarding
  • New ZopNight: one-click storage-lifecycle remediation across all three clouds: move GCS buckets, S3 buckets, and Azure storage accounts to colder tiers at 30 and 90 days. Transitions are lossless (no delete), with an Adopt / Replace toggle and full Restore
  • New ZopDay: a live Observability tab on the service drawer: replica count, restarts, per-replica CPU / memory, HPA targets, scoped events, and a multi-replica log button, with progressive loading per card

Get release notes in your inbox.

We ship every two weeks. Subscribe to receive a short summary of what's new, no marketing, just the changelog in email form.

Subscribe →

Stop watching the waste.
Start cutting it.

See. Find. Fix. Automatic.

Connect your first cloud account in under 5 minutes. See your first remediation in under 7. No credit card required.

weekly releases
every change documented
zero breaking changes since v1.0
Multi-cloud automation· Production-ready in 30 min· SOC 2 · ISO 27001· 20–60% off the bill, first month· 4 platforms · 1 console· Multi-cloud automation· Production-ready in 30 min· SOC 2 · ISO 27001· 20–60% off the bill, first month· 4 platforms · 1 console·