Skip to content

Observability

Observability API client for kdb Insights Enterprise.

Use ObservabilityService to stream logs, query Kubernetes events, and list running workload pods in a deployed kdb Insights Enterprise namespace.

Example

Fetch and filter logs:

from kxi.observability import ObservabilityService

obs = ObservabilityService(host="https://<host>")

# Fetch the last 100 log lines for a namespace
logs = obs.get_logs(namespace="my-namespace", tailLines=100)
for log in logs:
    print(log.level, log.message)
Example

Stream logs continuously and inspect workload health:

# Stream logs until Ctrl+C
obs.watch_logs(namespace="my-namespace", pretty=True)

# Query Kubernetes events for a specific workload
events = obs.get_events(namespace="my-namespace", workload="assembly1")
for event in events:
    print(event.reason, event.message)

# List all running pods
pods = obs.get_pods(namespace="my-namespace")
for pod in pods:
    print(pod.pod, pod.phase, pod.containers)

Classes:

Functions:

ObservabilityLogLevel

Bases: str, Enum

Log severity levels used to filter or classify log messages.

Pass as the level parameter to ObservabilityService.get_logs.

Attributes:

  • TRACE – Most verbose level, below DEBUG.
  • DEBUG – Detailed diagnostic information.
  • INFO – Standard informational messages.
  • WARN – Warning conditions that don't cause failures.
  • WARNING – Alias for WARN.
  • ERROR – Error conditions that may cause partial failure.
  • FATAL – Critical failure — process cannot continue.

ObservabilityLogMsg

Bases: BaseModel

Structured log message returned by the Observability API.

All fields are optional. Unrecognised API fields are preserved via extra="allow".

Attributes:

  • level (ObservabilityLogLevel | None) – Severity level.
  • message (str | None) – Log message text.
  • time (str | None) – ISO 8601 timestamp.
  • pod (str | None) – Pod name that emitted the log.
  • container (str | None) – Container name within the pod.
  • component (str | None) – Logical component (e.g. pipeline name).
  • service (str | None) – Service identifier.
  • host (str | None) – Host or node name.
  • corr (str | None) – Correlation ID for request tracing.
  • instance (str | None) – Instance identifier.
  • source (str | None) – Source of the log entry.
  • status (int | None) – HTTP status code, when applicable.
  • method (str | None) – HTTP method, when applicable.
  • uri (str | None) – Request URI, when applicable.
  • user (str | None) – Authenticated user, when applicable.

ObservabilityService

Bases: ApiClient

REST client for the kdb Insights Enterprise Observability API.

Provides access to logs, Kubernetes events, and workload pod information. Connects to the /observe service path on the kdb Insights host.

Example 
from kxi.observability import ObservabilityService

obs = ObservabilityService(host="https://<host>")

logs = obs.get_logs(namespace="my-namespace", tailLines=50)
pods = obs.get_pods(namespace="my-namespace")

See Connection Parameters for shared constructor arguments.

Functions:

  • a_get_logs – Asynchronously yield log messages as they arrive.
  • async_watch_logs – Asynchronously stream logs until the task is cancelled.
  • get_events – Fetch Kubernetes events for a namespace.
  • get_logs – Fetch logs from a namespace.
  • get_pods – Fetch the list of running pods in a namespace.
  • watch_logs – Stream logs continuously until interrupted with Ctrl+C.

a_get_logs

a_get_logs(**params)

Asynchronously yield log messages as they arrive.

When follow=False, fetches current logs once and returns. When follow=True (default), streams logs continuously.

Parameters:

  • *params* – Filtering parameters forwarded to the log API (see get_logs).

Yields:

async_watch_logs

async_watch_logs(**params)

Asynchronously stream logs until the task is cancelled.

Parameters:

  • *params* – Filtering parameters forwarded to the log API (see get_logs).

get_events

get_events(**params)

Fetch Kubernetes events for a namespace.

Parameters:

  • namespace – Target Kubernetes namespace.
  • pod – Exact pod name to retrieve events for.
  • package – Filter pods associated with a specific package.
  • workload – Limit results to components related to this workload.
  • label – Comma-separated key-value label selectors.
  • pretty – Render response with indentation.
  • *params* – Additional keyword arguments passed to the events API.

Returns:

get_logs

get_logs(**params)

Fetch logs from a namespace.

Parameters:

  • namespace – Target Kubernetes namespace.
  • pod – Filter by pod name.
  • container – Filter by container name.
  • workload – Limit results to components related to this workload.
  • label – Comma-separated key-value label selectors.
  • component – Regex applied to the workload name.
  • level – Minimum log level to return (see ObservabilityLogLevel).
  • logComponent – Filter by the log message component field.
  • search – Case-insensitive substring search on the message field.
  • corr – Filter by correlation ID.
  • sinceSeconds – Return logs from this many seconds ago.
  • sinceTime – RFC 3339 timestamp from which to return logs.
  • tailLines – Number of lines from the end of the log to return per container.
  • aggregate – Aggregate all log messages by log level.
  • byPod – Group results by pod name and log level (requires aggregate).
  • pretty – Pretty-print the response JSON.
  • *params* – Additional keyword arguments passed to the log API.

Returns:

get_pods

get_pods(**params)

Fetch the list of running pods in a namespace.

Parameters:

  • namespace – Target Kubernetes namespace.
  • *params* – Additional keyword arguments passed to the workloads API.

Returns:

watch_logs

watch_logs(**params)

Stream logs continuously until interrupted with Ctrl+C.

Blocks the calling thread. Pass follow=False to fetch current logs once and exit.

Parameters:

  • *params* – Filtering parameters forwarded to the log API (see get_logs).

WorkloadEvent

Bases: BaseModel

Kubernetes event associated with a workload pod.

Attributes:

  • count (int | None) – Number of times the event has occurred.
  • first_time (str | None) – Timestamp of the first occurrence.
  • last_time (str | None) – Timestamp of the most recent occurrence.
  • message (str | None) – Human-readable event description.
  • name (str | None) – Event object name.
  • namespace (str | None) – Kubernetes namespace.
  • pod (str | None) – Name of the related pod.
  • reason (str | None) – Short machine-readable reason (e.g. "OOMKilling").
  • type (str | None) – Event type — "Normal" or "Warning".
  • workload (str | None) – Name of the related workload.

WorkloadPod

Bases: BaseModel

Pod information for a running workload.

Attributes:

  • containers (list[str]) – Names of containers running in the pod.
  • package (str | None) – Package deployed into this pod.
  • phase (str | None) – Kubernetes pod phase (e.g. "Running", "Pending").
  • pod (str | None) – Pod name.
  • workload (str | None) – Workload name this pod belongs to.

drop_nones

drop_nones(d)

Drop None values from a dict.

get_remote_logger_func

get_remote_logger_func(name=DEFAULT_OBS_LOG_NAME, pretty=False)

Get a pretty remote logger.

parse_time_stamp

parse_time_stamp(time)

Parse time stamp.

Back to top