OCI Observability Professional flashcards
OCI Observability Professional · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- Metric namespace
- A container that groups related metrics published by one source, such as oci_computeagent. Queries must name the namespace before naming a metric key inside it.
- Metric dimension
- A key/value attribute attached to each data point, such as resourceId or availabilityDomain. Dimensions let you filter and group query results by resource or custom tag.
- MQL alarm query
- A Monitoring Query Language expression combining a metric, statistic, interval, and optional grouping. The alarm compares the evaluated result against a threshold over a rolling window.
- Alarm FIRING state
- The state entered when the evaluated metric statistic breaches the threshold for the configured window. It triggers notification delivery to the alarm's configured topic.
- Alarm RESET state
- The state entered when a previously firing metric returns inside the threshold. It signals recovery and can send its own notification so responders know the incident cleared.
- Basic monitoring resolution
- Standard service metrics are published at one-minute granularity without any agent. Finer resolution requires agent-based or custom metric publication, which has its own limits.
- Health check versus metric
- A health check reports reachability and availability of an endpoint, while metrics report performance and saturation. A resource can be healthy yet show degrading latency metrics.
- Log group
- A logical container that holds one or more logs. Access policies and search scoping apply at this level, so grouping by application or environment simplifies authorization.
- Agent log configuration
- A definition specifying the source file path, the parser to apply, and the destination log. Multiple configurations can run on a single unified monitoring agent.
- Service log
- A log emitted directly by an OCI service, such as VCN flow logs or load balancer access logs. It is enabled per resource or compartment and needs no agent.
- Audit log
- A service log recording control-plane API activity against your tenancy. It is distinct from application and data-plane logs but can be routed to the same log group.
- Log parser
- The component that converts raw log text into structured fields, using formats such as syslog, apache, JSON, or a custom regular expression. Parsing enables field-based search.
- Log archival
- Moving logs past their retention window into Object Storage for long-term keeping. Archived logs are no longer searchable in the Logging service until restored or re-ingested.
- Custom log ingestion
- Uploading log entries through the Logging Ingestion API rather than an agent. Requires an endpoint, credentials, and a target log, and suits applications that push directly.
- APM trace
- The complete record of one request as it travels across services, composed of linked spans. It shows where time was spent and which component failed.
- APM span
- A single timed operation within a trace, with a kind such as server or client, attributes, and a parent reference. Spans nest to form the trace's call tree.
- APM domain
- The top-level isolation boundary for APM traces, spans, and configuration. It lives in a compartment and is referenced by agents and SDKs when uploading data.
- APM data key
- A public and private key pair that authenticates agents and SDKs to an APM domain. The private key authorizes uploads and must be stored securely.
- Auto-instrumentation
- Attaching an APM agent to a runtime, such as Java or .NET, so common frameworks are traced without code changes. Manual spans can be added for custom logic.
- Latency percentile
- A value such as p95 or p99 showing the response time under which that share of requests fall. Percentiles expose tail latency that averages hide.
- Synthetic monitoring
- Scripted browser or REST checks run from multiple vantage points on a schedule. They produce availability and latency metrics that can drive alarms.
- Browser agent
- An APM component embedded in web pages that reports real user page-load and interaction timing. It complements server-side traces with client-side performance data.
- Service connector
- A managed pipeline with a source, optional task, and target that continuously moves data between OCI services. It requires IAM policies for source read and target write.
- Connector task
- An in-flight transformation inside a service connector, such as a log filter query or a Functions invocation. Tasks shape or enrich data before it reaches the target.
- Connector service principal
- The identity a service connector uses when reading from its source and writing to its target. Missing target policies cause silent delivery failures.
- Notification topic
- A named channel that alarm rules and other publishers send messages to. Subscriptions attached to the topic determine which endpoints actually receive them.
- Subscription confirmation
- The step where a subscription recipient proves ownership, such as clicking an email link. Unconfirmed subscriptions exist but receive no messages.
- Alarm severity
- The classification attached to an alarm, one of informational, warning, or critical. It is included in notifications and helps consumers route and prioritize responses.
- Repeat notification
- A setting that re-sends an alarm message at a fixed interval while the alarm stays firing. It keeps responders aware but can add noise during long incidents.
- Alarm message format
- A template for notification content that can embed the metric value, dimensions, and state. It determines what on-call readers see without opening the console.
- Streaming source
- A service connector source that reads records from a stream and forwards them to a target such as Object Storage. It supports high-volume event pipelines.
- Monitoring metric source
- A service connector can use Monitoring metric data points as its source, enabling periodic export of metrics to Object Storage or another target for long-term analysis.