Class PrometheusTimeseries

python/ray/_common/test_utils.py:430–445

A collection of timeseries from multiple addresses. Each timeseries is a collection of samples with the same metric name and labels. Concretely: - components_dict: a dictionary of addresses to the Component labels - metric_descriptors: a dictionary of metric names to the Metric object

@dataclass
class PrometheusTimeseries:
    """A collection of timeseries from multiple addresses. Each timeseries is a
    collection of samples with the same metric name and labels. Concretely:
    - components_dict: a dictionary of addresses to the Component labels
    - metric_descriptors: a dictionary of metric names to the Metric object
    - metric_samples: the latest value of each label
    """

    components_dict: Dict[str, Set[str]] = field(default_factory=dict)
    metric_descriptors: Dict[str, "Metric"] = field(default_factory=dict)
    metric_samples: Dict[frozenset, "Sample"] = field(default_factory=dict)

    def flush(self):
        self.components_dict.clear()
        self.metric_descriptors.clear()
        self.metric_samples.clear()


def fetch_raw_prometheus(prom_addresses, timeout=None):