You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
To give better visibility into the costs of taskbroker/workers we should record usage in finer detail.
Workers
Workers should already be measuring task execution time. We should use sentry.usageaccountant to record time spent on tasks per app_feature. We'll need a way to map between namespaces and app_features. Perhaps that could be metadata that is included in the TaskNamespaces.
Brokers
Brokers could periodically emit task counts for each namespace they are handling. Again we'd need a map between namespaces and app_features. The https://github.com/getsentry/rust-usage-accountant library can be used to record and publish usage results.
The text was updated successfully, but these errors were encountered:
app_feature: the same as the corresponding celery task
system: task_broker
subsystem: broker / worker.
As brokers ingest messages from Kafka we could generate counts of activations by app_feature and emit those as kafka messages using usage-accountant. This implies that there is a configuration map of namespace : app_feature within broker configuration.
To give better visibility into the costs of taskbroker/workers we should record usage in finer detail.
Workers
Workers should already be measuring task execution time. We should use
sentry.usageaccountant
to record time spent on tasks per app_feature. We'll need a way to map between namespaces and app_features. Perhaps that could be metadata that is included in the TaskNamespaces.Brokers
Brokers could periodically emit task counts for each namespace they are handling. Again we'd need a map between namespaces and app_features. The https://github.com/getsentry/rust-usage-accountant library can be used to record and publish usage results.
The text was updated successfully, but these errors were encountered: