opmon/firefox-background-update.toml

# Monitoring of Firefox background update process.

[project]

name = "Firefox Background Update"
platform = "firefox_desktop"
xaxis = "submission_date"
start_date = "2022-01-01"
# We want continuous monitoring of this data:
# end_date =
skip_default_metrics = true

metrics = [
    "client_volume",
    "ping_volume",
    "success_rate",
    "exception_rate",
    # It turns out that the success rate is ~1.0 and the exception rate ~0.0, so
    # the exception volume is more useful.
    "exception_volume",
    "crash_volume",
    # TODO: "exit_state_rate",
]

[project.population]

data_source = "background_update"
monitor_entire_population = true
dimensions = ["normalized_channel", "windows_version"]

[dimensions.windows_version]
data_source = "background_update"
select_expression = "windows_version_info"
friendly_name = "Windows Version"
description = "Windows version, like 'Windows 11', 'Windows 10', etc.  See https://github.com/mozilla/bigquery-etl/blob/generated-sql/sql/mozfun/norm/windows_version_info/udf.sql."

[dimensions.normalized_channel]
data_source = "background_update"
select_expression = "normalized_channel"
friendly_name = "Channel"
description = "Release channel, like 'release', 'beta', 'nightly', 'esr', 'devedition'."

[data_sources]

[data_sources.background_update]
from_expression = """
    ( SELECT * FROM (
        SELECT
            DATE(submission_timestamp) AS submission_date,
            -- We don't care about builds and versions at this point.
            NULL AS build_id,
            client_info.client_id AS client_id,
            metrics.uuid.background_update_client_id AS legacy_client_id,
            mozfun.norm.windows_version_info(
              "Windows_NT",
              CONCAT(SPLIT(normalized_os_version, ".")[SAFE_OFFSET(0)], ".", SPLIT(normalized_os_version, ".")[SAFE_OFFSET(1)]),
              COALESCE(client_info.windows_build_number, 0)
              -- SAFE_CAST(SPLIT(normalized_os_version, ".")[SAFE_OFFSET(2)] AS INT64)
              ) AS windows_version_info,
            normalized_channel AS normalized_channel,

            -- It's more convenient to always have these defined, and always as integers (0 or 1).
            CAST(COALESCE(metrics.boolean.background_update_exit_code_success, FALSE) AS INT64) AS exit_code_success,
            CAST(COALESCE(metrics.boolean.background_update_exit_code_exception, FALSE) AS INT64) AS exit_code_exception,
        FROM
            mozdata.firefox_desktop_background_update.background_update
    )
        WHERE
            -- Some clients are badly configured.
            windows_version_info IS NOT NULL
            -- Ancient versions of Windows are not relevant.
            AND windows_version_info NOT IN ('Windows 95', 'Windows 98', 'Windows NT 4.0', 'Windows 2000', 'Windows XP', 'Windows Vista')
    )
"""
submission_date_column = "submission_date"
build_id_column = "build_id"
client_id_column = "legacy_client_id"

[data_sources.background_update_crashes]
from_expression = """
    ( SELECT * FROM (
        SELECT
            DATE(submission_timestamp) AS submission_date,
            -- We don't care about builds and versions at this point.
            NULL AS build_id,
            -- This must match the client ID of the `background_update` data source!
            client_id,
            mozfun.norm.windows_version_info(
              "Windows_NT",
              environment.system.os.version,
              SAFE_CAST(environment.system.os.windows_build_number AS INT64)
              ) AS windows_version_info,
            normalized_channel AS normalized_channel,
        FROM
            mozdata.telemetry.crash
        WHERE
            payload.metadata.background_task_name = 'backgroundupdate'
    )
        WHERE
            -- Some clients are badly configured.
            windows_version_info IS NOT NULL
            -- Ancient versions of Windows are not relevant.
            AND windows_version_info NOT IN ('Windows 95', 'Windows 98', 'Windows NT 4.0', 'Windows 2000', 'Windows XP', 'Windows Vista')
    )
"""
submission_date_column = "submission_date"
build_id_column = "build_id"
client_id_column = "client_id"

[metrics]

[metrics.ping_volume]
data_source = "background_update"
select_expression = "COUNT(*)"
type = "scalar"

[metrics.ping_volume.statistics]
sum = {}
# TODO: one week and one month smoothed?

[metrics.client_volume]
data_source = "background_update"
select_expression = "COUNT(DISTINCT legacy_client_id)"
type = "scalar"

[metrics.client_volume.statistics]
sum = {}
# TODO: one week and one month smoothed?

[metrics.success_rate]
data_source = "background_update"
select_expression = "AVG(exit_code_success)"
type = "scalar"

[metrics.success_rate.statistics]
mean = {}

[metrics.exception_rate]
data_source = "background_update"
select_expression = "AVG(exit_code_exception)"
type = "scalar"

[metrics.exception_rate.statistics]
mean = {}

[metrics.exception_volume]
data_source = "background_update"
select_expression = "SUM(exit_code_exception)"
type = "scalar"

[metrics.exception_volume.statistics]
sum = {}

[metrics.crash_volume]
data_source = "background_update_crashes"
select_expression = "COUNT(*)"
type = "scalar"

[metrics.crash_volume.statistics]
sum = {}