SystemLogs

Package: flyteplugins.union.remote

Logs of the Union system components running on a cluster’s dataplane.

Selected by Kubernetes label rather than by any Flyte object.

Parameters

        
    
def SystemLogs()

Methods

Method	Description
`create_viewer()`	Stream a cluster’s system logs to the console.
`tail()`	Stream the system logs of a cluster’s dataplane components.

create_viewer()

        
    
def create_viewer(
    cluster_name: str,
    namespace: str,
    application: Application | None = 1,
    pod_label_selector: str | None = None,
    start_time: datetime | None = None,
    node_name: str | None = None,
    source: logs_pb2.LogsSource = 0,
    max_lines: int = 30,
    show_ts: bool = False,
    raw: bool = True,
)

Stream a cluster’s system logs to the console.

The remaining arguments select the stream and are passed to tail.

Parameter	Type	Description
`cluster_name`	`str`
`namespace`	`str`
`application`	`Application \| None`
`pod_label_selector`	`str \| None`
`start_time`	`datetime \| None`
`node_name`	`str \| None`
`source`	`logs_pb2.LogsSource`
`max_lines`	`int`	Lines kept in view, when not raw.
`show_ts`	`bool`	Whether to prefix each line with its timestamp.
`raw`	`bool`	Print lines as they arrive, scrolling the terminal. When False, they are shown in an auto-scrolling box of `max_lines` lines.

tail()

This method can be called both synchronously or asynchronously.

Default invocation is sync and will block. To call it asynchronously, use the function .aio() on the method name itself, e.g.,: result = await SystemLogs.tail.aio().

        
    
def tail(
    cls,
    cluster_name: str,
    namespace: str,
    application: Application | None = 1,
    pod_label_selector: str | None = None,
    start_time: datetime | None = None,
    node_name: str | None = None,
    source: logs_pb2.LogsSource = 0,
) -> AsyncGenerator[logs_pb2.LogLine, None]

Stream the system logs of a cluster’s dataplane components.

Ends when the dataplane closes the stream; there is no reconnect. Lines are prefixed with their pod once more than one pod is in the stream.

Parameter	Type	Description
`cls`
`cluster_name`	`str`	Name of the cluster whose dataplane to read from.
`namespace`	`str`	Kubernetes namespace the components run in.
`application`	`Application \| None`	Which component to tail. Mutually exclusive with `pod_label_selector`; pass `None` when using one.
`pod_label_selector`	`str \| None`	Kubernetes label selector to pick pods with, for components the `Application` enum does not name.
`start_time`	`datetime \| None`	Earliest time to read logs from. Defaults to the start of the retained logs.
`node_name`	`str \| None`	Only stream logs of pods running on this Kubernetes node.
`source`	`logs_pb2.LogsSource`	Whether to read live logs, persisted logs, or (the default) live logs falling back to persisted when no live pods match.