OpenStack gets noisy fast. Atmosphere cuts through it with real metrics, real logs, and real alerts, so teams can fix issues before users notice.
Operating OpenStack at scale means tracking dozens of interdependent services across compute, storage, and networking. Teams need to monitor virtual machine (VM) provisioning, API responsiveness, network agent behavior, storage pool health, and SO MUCH more. Without the right visibility, issues can escalate before anyone notices.
Atmosphere includes an integrated observability stack that covers real-time monitoring, log aggregation, tracing, and alerting. These tools give operators and SREs what they need to troubleshoot quickly, maintain reliability, and scale confidently.
Metrics and Dashboards That Reflect Operational Reality
Atmosphere uses Prometheus to capture metrics across OpenStack and Kubernetes layers. This includes everything from nova-scheduler queue length to Ceph pool rebalance activity.
Grafana dashboards are preconfigured with views for Nova, Neutron, Glance, Keystone, and Ceph. These dashboards track resource saturation, error rates, service availability, and usage patterns. Each chart has been built based on what operators need during active troubleshooting, and not generic infrastructure templates.
Teams can also define custom metrics to support internal SLAs or tenant-specific KPIs.
Centralized Logging for Faster Debugging
OpenStack issues often require context across multiple services. A failed VM launch may touch Nova, Glance, Cinder, and Neutron. Atmosphere uses centralized logging tools like Elasticsearch or Loki to collect logs across all layers.
Logs are indexed and structured so that users can query based on instance UUIDs, tenant IDs, service names, or error patterns. Pre-built queries make it easier to identify known issues, such as repeated API timeouts or volume attach retries.
Log retention policies are configurable, allowing operators to control disk usage while retaining historical data for audits or long-term troubleshooting.
Tracing Slowdowns and Failures Across Services
Distributed requests often span multiple OpenStack components. Tracing tools like OpenTelemetry and Jaeger are integrated into Atmosphere so teams can follow a request as it moves through Horizon, Keystone, Nova, and downstream storage or networking.
Traces are useful when dealing with API slowness, resource creation delays, or unresponsive endpoints. They show where the time is spent, what services introduce latency, and what downstream failures block progress.
