@kapa.ai ga4 backoff isnt long enough, keeps hitting api limits and then times out the job and does 2nd attempt, starting over

Thinking completed Based on the provided documentation and community discussions, this issue typically occurs when the Google Analytics Data API (GA4) exhausts its hourly or daily quota tokens (such as

RESOURCE_EXHAUSTED

/

"Exhausted property tokens for a project per hour"

), while the standard retry backoff duration is too short for the quota window to reset, causing retries to fail and the job to restart from the beginning [Forum Post: Exhausted property tokens error on GA4 initial run; Documentation Platform > Understanding-airbyte > Workloads & jobs>]. Here are the recommended ways to mitigate this in Airbyte: 1. Increase the Date Request Interval (window_in_days) By default, the connector queries data in 1-day chunks (

window_in_days: 1

). Pulling small date ranges repeatedly in rapid succession consumes significant hourly token quotas (each request costs multiple tokens). • In the GA4 source settings, increase Data Request Interval (Days) (

window_in_days

) to a larger window (e.g.,

30

,

60

, or higher) [manifest.yaml spec > connection_specification > properties>; Forum Post: Exhausted property tokens error on GA4 initial run]. • A larger range makes fewer total API calls to pull the same volume of data, significantly cutting down on token consumption [Forum Post: Exhausted property tokens error on GA4 initial run]. (Note: Larger windows increase the chance of GA4 sampling data, so balance this against your accuracy needs). 2. Verify or Set Your Subscription Plan/Tier The connector enforces a local

api_budget

policy based on your selected tier [manifest.yaml]: • Standard Property: Rate-limited locally to 10 requests/second. • Analytics 360 Property: Allows up to 50 requests/second. Ensure the Subscription Plan/Tier setting matches your actual property tier [manifest.yaml spec > connection_specification > properties>]. If you are on a Standard Property, running high concurrency or selecting Analytics 360 will quickly trigger

429

quota exhaustion errors [manifest.yaml]. 3. Sync Fewer Streams at a Time If multiple high-cardinality streams are running concurrently, they compete for the same property hourly token pool. To get past the initial historical backfill without exhausting quotas: • Temporarily disable optional streams and sync the essential ones first. • Once the initial sync succeeds and state bookmarks are saved, enable remaining streams. Subsequent incremental runs consume far fewer tokens. 4. Quota Limitations Unlike many other Google APIs, Google Analytics Data API quotas cannot be self-service increased directly from the Google Cloud Platform (GCP) console [source-google-analytics-data-api: Unique Behaviors]. Quota increases require coordination through Google support or your reseller [source-google-analytics-data-api: Unique Behaviors]. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.