Large number of GOAWAY responses from APNS over the last week

We have been seeing a large increase of errors since March 5, 2024 starting around 8pm UTC.

Example error response is

http2: server sent GOAWAY and closed the connection; LastStreamID=2147483647, ErrCode=PROTOCOL_ERROR, debug="Stream 169 does not exist"

We've been sending notifications the same way for about 8 months now, and a very similar way for over 2 years without this issue.

Currently getting about an 8% failure rate on requests. Sending about 20,000,000 notifications a day.

Has something changed that we need to modify our process for?

Large number of GOAWAY responses from APNS over the last week
 
 
Q