@kapa.ai my amplitude connector keeps getting stuck. Any suggestions?

Thinking completed Here are some suggestions for when your Amplitude connector gets stuck: 1. Large Events Exports (Most Common Cause) The Events stream can get stuck or fail when export windows exceed 4 GB or time out. The connector automatically splits the time window in half and retries, but there are limits: • With a Request Time Range of up to 1024 hours, splitting can reach a 1-hour window floor. • Above 1024 hours, splitting stops at windows of 8–9 hours, which may still be too large. • If a 1-hour window still exceeds 4 GB, the sync fails with a transient error and must be retried. What to do: • Lower your

Request Time Range

(default is 24 hours). If most syncs need to split, a smaller value reduces the number of extra failing requests. • If a single hour of data exceeds 4 GB, you'll need to use Amplitude's Amazon S3 export instead. [Performance considerations] 2. Rate Limits The Dashboard REST API enforces: • 108,000 cost per hour • 1,000 cost per 5-minute burst window • Max 5 concurrent requests The connector handles these automatically, but if you're hitting limits that aren't being retried, consider filing an issue. [Performance considerations] 3. Active Users Stream Issues If the Active Users stream is causing problems, try setting

active_users_group_by_country

to

false

in your connector configuration. According to Amplitude's documentation, grouping by country is optional and can sometimes cause fetch issues. [manifest.yaml] 4. Use Append + Deduped Sync Mode If you're retrying after failures, use Append + Deduped sync mode for the Events stream. This removes duplicate records by

uuid

that may be re-read during retries. [Performance considerations] Could you share more details about which stream is getting stuck and any error messages you're seeing? That would help narrow down the cause further. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.