Accounty Activity API is sending us events sometimes 30 seconds to almost a whole minute late!

[bobber205]

Did some digging in our logs today and found something a bit concerning:

I am logging to a database events we receive and what time the user sent it (according to twitter’s payload) and the time on the server.

This is about 70% < 1 second from each other. Sometimes there is a whole second’s worth of difference.

But today for periods of time (about 20 seconds) we are receiving them 30 seconds to almost a minute in a few cases behind.

This makes our bot seem poorly made or optimized.
Is more guaranteed timely delivery part of the “reliability” that is listed under paid access to this service (like having more than X webhook URLS is for instance)

Thanks!

[andypiper]

This sounds like a latency issue related to system load. There can be some ebbs and flows over time, but I’ve not really seen this as a consistent or constant issue. Are you seeing this happening frequently?

The reliability features are more related to retries, rather than timeliness of delivery.

[bobber205]

When Twitter is under load do paying customers (premium and enterprise) have prioritized delivery?
Is the load account based or twitter system wide?

[KyleW]

Today, there is no difference in our infrastructure in this way, but this is feedback we are listening to and we may be able to help improve the experience for paying customers in the future.

There is no intention to add latency on these events, as @andypiper mentioned, this is usually just based on our system performance at any point in time.