To Reproduce Steps to reproduce the behavior:

1. Have CloudKitty disabled
2. Have a Prometheus with retention higher than 7 days.
3. Scrape data into Prometheus and wait for over a week
4. Make sure it's currently after 7th day of the current month
5. Enable CloudKitty
6. See errors in cloudkitty-proc's logs similar to:
   entry for stream '{service="cloudkitty", service_name="cloudkitty"}' has timestamp too old: 2025-10-04T14:50:00Z

Expected behavior

• No error logs should show up. CloudKitty should work

Bug impact

• CloudKitty reports might not be accurate for the current month

Known workaround

• Have Prometheus retention less than 7 days
• Or delete data older than 7 days from Prometheus before starting CloudKitty
• Or wait until next month