Skip to main content

Storage, Networking & Events

Alongside workloads and node groups, each cluster's detail page includes three resource views for inspecting the cluster's storage, networking, and event stream. All three are read-only and available on every cluster — Atmosly-provisioned, cloud-imported, and token-imported (including read-only imports) — with no extra permission required.

You reach them from the cluster's left navigation: Storage, Ingress & Service Accounts, and Cluster Events.

Storage​

The Storage view breaks down into several sub-views, with a shared filter bar across the top — filter by namespace and status, toggle Orphaned only to surface unmounted claims, and refresh on demand.

Overview​

A summary of the cluster's storage health:

  • Stat cards — counts for PVCs (Bound / Pending / Lost), PVs (Available / Bound / Released / Failed), Storage Classes (with the default highlighted), and Volume Attachments (Active / Failed).
  • Storage Health Score — a radial score with a status breakdown (PV status, PVC binding). Guide: Healthy ≥ 90%, Warning 70–89%, Critical < 70%.
  • Insights — auto-generated signals, such as pending PVCs, or PVs left Released with a Retain reclaim policy for over 14 days (worth cleaning up to avoid ongoing storage cost).

Persistent Volumes (PVs)​

Columns: Name, Status (Available / Bound / Released / Failed), Capacity, Access Modes, StorageClass, Claim (namespace/name), Age. Expanding a row shows reclaim policy, volume mode, node affinity, the claim reference, and labels.

Persistent Volume Claims (PVCs)​

Columns: Name, Status (Bound / Pending / Lost), Namespace, Requested capacity, Access / Mode, StorageClass, PV, Mounted By (the workload using it, or "Not mounted"). Pending claims show the binding error inline.

Storage Classes​

Columns: Name (default class tagged), Provisioner, Reclaim (Delete / Retain), Expansion (allowed or not), and whether it's In Use. Expanding shows volume-binding mode, parameters, and the PVCs using the class.

Volume Attachments​

Columns: Name, Status (Attached / Detached / Failed), Node, Attached, PV Name. Expanding shows the device path, attach/detach times, and any attach/detach errors.

Ingress & Service Accounts​

The networking view has two tabs:

Service Accounts​

Columns: Name, Namespace, Secrets (count, with names on hover), Image Pull Secrets (count), Age. Expanding a row shows UID, resource version, creation time, labels, and annotations.

Ingress & Load Balancers​

Columns: Name, Namespace, Type (LoadBalancer / Ingress / Service), External IP / Host, Ports (e.g., 80/TCP), Age. Use this to see which services are exposed externally and on what addresses and ports.

Cluster Events​

The Cluster Events view streams the cluster's Kubernetes events from the last 24 hours.

Columns: Namespace, Type (Normal / Warning), Object (kind and name), Reason, Count (highlighted as it climbs), First Seen (relative time). Filter by namespace and by event type (Normal / Warning), then search by object.

tip

The Cluster Events item in the left navigation carries a badge counting Warning events from the last 24 hours — a quick signal that something in the cluster needs attention. The SRE Agent turns recurring event patterns into actionable issues with remediation.