Closed zaphod534 closed 1 year ago
I'm getting this too on 2.0.0-rc0
Just tested this on a stock project and it is reproducible
Issue exists in a production app for 2.0.0-rc.0, sending to ~1k devices.
I thought it was a batching issue on my part then noticed the issue ticket.
After further testing, some of these timeouts are actually failing and FCM messages are not being delivered
Merged #217. I wonder how I missed that in the 2.0 refactor. Let me address some of the issues in Kadabra and see if it needs a new version before cutting a new release candidate here.
Thanks for everyone's help on this. I've been out of the loop the past month or so.
@hpopp were you able to cut a new 2.0 release candidate with the fix? Thanks.
Published 2.0.0-rc.1
. I also updated goth to 1.3 now that it's out of RC https://github.com/codedge-llc/pigeon/pull/224.
So I need to address #221 and do some minor cleanup on my end before I consider us good for 2.0 official release. It seems like with how busy I've been over the last year, time keeps slipping by. In a perfect world I could work on nothing but my open source packages, but of course that'll never happen 😅
Environment
Current behavior
After a while, the library stop processing the notifications response and give the
:timeout
error cause, even though the notification is successfully sent to the FCM / APNS servers and the response is successfully received.After digging into the problem, I noticed that the issue happens when the HTTP2
stream_id
numbering restart from 1, while the library keeps incrementing it, so every reference for any subsequent notification on that worker is lost.Steps to reproduce:
Expected behavior
The library should process the notification response.