pangolier 0.2.1

prometheus query builder

Pangolier

prometheus query builder

install

pip install pangolier

usage

simple case

For a metric with filters:

from pangolier.metrics import Metric

print(Metric('http_requests_total').filter(
    job='prometheus',
    group='canary'
).to_str())

output:

http_requests_total{job="prometheus", group="canary"}

pretty output

Add pretty=True in to_str for better readability:

from pangolier.metrics import Metric

print(Metric('http_requests_total').filter(
    job='prometheus',
    group='canary'
).to_str(pretty=True))

output:

http_requests_total{
    job="prometheus",
    group="canary"
}

I will always use pretty=True in rest of this document.

functions

Prometheus functions can be built by name. For example:

from pangolier.functions import function

abs = function('abs')
print(abs(Metric('http_requests_total')).to_str(pretty=True))

output:

abs(
    http_requests_total
)

range_function should be used for functions accept a range-vector.

from pangolier.functions import range_function

rate = range_function('rate')
print(rate(Metric('http_requests_total'), timespan='5m').to_str(pretty=True))

output:

rate(
    http_requests_total[5m]
)

aggregation_operator shoule be used for aggregation operators:

from pangolier.functions import aggregation_operator

sum_ = aggregation_operator('sum')
print(sum_(
    Metric('http_requests_total'),
    by=['job', 'group'],
).to_str(pretty=True))

output:

sum by(
    job, group
)(
    http_requests_total
)

combine them all together:

histogram_quantile = function('histogram_quantile')
rate = range_function('rate')
sum_ = aggregation_operator('sum')

print(histogram_quantile(
    0.9,
    sum_(
        rate(
            Metric('http_request_duration_seconds_bucket'),
            timespan='5m',
        ),
        by=['le']
    )
).to_str(pretty=True))

output:

histogram_quantile(
    0.9,
    sum by(
        le
    )(
        rate(
            http_request_duration_seconds_bucket[5m]
        )
    )
)

binary operators

support following binary operators:

• + (addition)
• - (subtraction)
• * (multiplication)
• / (division)
• % (modulo)
• ^ (power/exponentiation)

For example, divide one metric with another:

from pangolier.metrics import Metric
from pangolier.functions import range_function

rate = range_function('rate')
print((
    rate(
        Metric('foo').filter(
            group='canary'
        ),
        timespan='5m'
    ) / rate(
        Metric('bar').filter(
            group='canary'
        ),
        timespan='5m'
    )
).to_str(pretty=True))

output:

rate(
    foo{
        group="canary"
    }[5m]
) / rate(
    bar{
        group="canary"
    }[5m]
)

For operation with modifier:

from pangolier.metrics import Metric, BinOp, GroupLeft

print(BinOp(
    '*',
    Metric('foo'),
    Metric('bar'),
    on=['interface', 'job'],
    group=GroupLeft('node', 'resource'),
).to_str(pretty=True))

output:

foo * on(
    interface, job
) group_left(
    node, resource
) bar

about name

Pangolier