fix(backend): clean up event batches stuck for a long time

[detj]

Summary

This PR cleans up event batches that are stuck in a pending state for a long time. This prevents certain situations where ingestion can get into a 429 Too Many Requests loop and hence stall ingestion.

Note

Only stale event batches having the same id as the previous batch gets removed. So, other event batches are not affected from this change.

Tasks

• Clean up event batches that are pending for a long time

Flow

The updated duplicate event batch check flow is as follows.

flowchart TB
	id1(Start) --> id2(Fetch request with 'msr-req-id')
	id2 --> id3{Is status pending?}
	id3 --y--> id4{Has 'created_at' passed timeout threshold?}
	id3 --n--> id5(Return '202 Accepted' known event request)
	id5 --> stop(Stop)
	id4 --y--> del(Delete pending event request)
	id4 --n--> tooMany(Return '429 Too Many Requests' response)
	tooMany --> stop
	del --> proceed(Proceed with regular ingestion flow)
	proceed --> stop

Loading

See also

• fixes Clean up all stuck event batch(s) in pending state #1769

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
measure-dashboard	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jan 23, 2025 3:56pm