Closed datdojp closed 3 years ago
Yes, it happens from time to time. I switched from ConnectionPool to 1 connection per worker (create new connection, use it to send notifiactions, close it) and it works.
@datdojp - I started the Apple developer forums thread you mentioned. I called into support (1-800-692-7753) and they said that not enough people had reported the issue to them so they do not believe it is an issue on their side
Yes, it happens from time to time.
@maciejkowalski I'm experiencing it consistently now - I have not been able to send a successful single push notification for any of our apps. Are you currently able to send pushes successfully on your end?
All info is in https://github.com/ostinelli/apnotic/issues/68.
TL;DR: apple changed their HTTP2 implementation and started enforcing headers ordering, at first only in the development environment. This has already been fixed in net-http2
2 weeks ago. Today, they started enforcing the same thing in production. Just upgrade apnotic
& net-http2
. Note that you now will need to add a block to your connection pool if you use it, i.e.
APNOTIC_POOL = Apnotic::ConnectionPool.new({
cert_path: Rails.root.join("config", "certs", "apns_certificate.pem"),
cert_pass: "mypass"
}, size: 5) do |connection|
connection.on(:error) { |exception| puts "Exception has been raised: #{exception}" }
end
Check README for info.
Thank you @ostinelli !
@maciejkowalski I'm experiencing it consistently now - I have not been able to send a successful single push notification for any of our apps. Are you currently able to send pushes successfully on your end?
Yes. 95% of them are received instantly, some of them are later (minutes). Sometimes I receive notification longly (hours) after Sidekiq worker done its job.
I upgraded apnotic
and everything works fine now. So many thanks for all your support.
By the way, I wonder how we prevent this kind of problem in the future. I mean how we know that Apple released their changes so that we can adapt our system accordingly. Do we need to subscribe to a forums, blog, channel or something like that?
Thanks a lot for this issue and update on net-http2.
This bug was a real pain for us since two days.
By the way, I wonder how we prevent this kind of problem in the future. I mean how we know that Apple released their changes so that we can adapt our system accordingly. Do we need to subscribe to a forums, blog, channel or something like that?
And I have the same question.
By the way, I wonder how we prevent this kind of problem in the future. I mean how we know that Apple released their changes so that we can adapt our system accordingly. Do we need to subscribe to a forums, blog, channel or something like that?
If you find a way let me know. @byronformwalt is also trying to see if anyone knows how to get this kind of information in advance.
Hi @ostinelli thank you for the great work on this library.
We have latest apnotic release & net-http2 release, but are still experiencing this issue periodically (lets say 4-5 times per day).
Gemfile.lock
apnotic (1.4.1)
connection_pool (~> 2.0)
net-http2 (>= 0.18, < 2)
Stacktrace
- 6 non-project frames
1
File "/app/vendor/ruby-2.4.4/lib/ruby/2.4.0/openssl/buffering.rb" line 181 in sysread_nonblock
2
File "/app/vendor/ruby-2.4.4/lib/ruby/2.4.0/openssl/buffering.rb" line 181 in read_nonblock
3
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 141 in block in socket_loop
4
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 138 in loop
5
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 138 in socket_loop
6
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 110 in block (2 levels) in ensure_open
SocketError: Socket was remotely closed
- 3 non-project frames
1
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 130 in callback_or_raise
2
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 115 in rescue in block (2 levels) in ensure_open
3
File "/app/vendor/bundle/ruby/2.4.0/gems/net-http2-0.18.0/lib/net-http2/client.rb" line 109 in block (2 levels) in ensure_open
Resending notifications with same payload/config does not cause the issue to occur again.
Any thoughts on how we could provide further diagnosis?
Thank you Matt
I don't think you can have more info than this one. See https://github.com/ostinelli/apnotic/issues/41#issue-203624698
Thanks for the update - let me know if you'd like me to move my comment to the other issue.
Hi there! We are using Apnotic in production since one week and it works fine, only that it fails sporadically with SocketError: Socket was remotely closed
I have seen the other issues but we are using the latest versions.
Gemfile.lock:
apnotic (1.5.0)
connection_pool (~> 2)
net-http2 (>= 0.18, < 2)
connection_pool (2.2.2)
net-http2 (0.18.2)
http-2 (~> 0.10.1)
We are also using sync pushes since we have a dedicated project.
Here is the backtrace:
SocketError: Socket was remotely closed
/usr/local/bundle/gems/net-http2-0.18.2/lib/net-http2/client.rb:130:in `callback_or_raise'
/usr/local/bundle/gems/net-http2-0.18.2/lib/net-http2/client.rb:115:in `rescue in block (2 levels) in ensure_open'
/usr/local/bundle/gems/net-http2-0.18.2/lib/net-http2/client.rb:109:in `block (2 levels) in ensure_open'
Thanks very much for the help, and awesome work with this library
Hi. APNS will remotely close connections for a variety of reasons (inactivity, server maintenance, server scaling). In my experience, even an inactive connection can remain open for many hours before being remotely closed. As far as I know, Apple does not specify the conditions that result in remotely closing connections. Can you post additional details about your case (e.g. stats on time before experiencing the remote connection is closed)?
On Jun 18, 2019, at 9:00 AM, Felipe Fava notifications@github.com wrote:
Hi there! We are using Apnotic in production since one week and it works fine, only that it fails sporadically with SocketError: Socket was remotely closed
I have seen the other issues but we are using the latest versions.
Gemfile.lock:
apnotic (1.5.0) connection_pool (~> 2) net-http2 (>= 0.18, < 2) connection_pool (2.2.2) net-http2 (0.18.2) http-2 (~> 0.10.1) We are also using sync pushes since we have a dedicated project.
Here is the backtrace:
SocketError: Socket was remotely closed /usr/local/bundle/gems/net-http2-0.18.2/lib/net-http2/client.rb:130:in
callback_or_raise' /usr/local/bundle/gems/net-http2-0.18.2/lib/net-http2/client.rb:115:in
rescue in block (2 levels) in ensure_open' /usr/local/bundle/gems/net-http2-0.18.2/lib/net-http2/client.rb:109:in `block (2 levels) in ensure_open' Thanks very much for the help, and awesome work with this library— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ostinelli/apnotic/issues/73?email_source=notifications&email_token=AAH2RJDO5YKNKKTNEBG6IZTP3EBDXA5CNFSM4FKUNX42YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODX7D2WQ#issuecomment-503201114, or mute the thread https://github.com/notifications/unsubscribe-auth/AAH2RJC4FWQVKWJVSI5JBKTP3EBDXANCNFSM4FKUNX4Q.
It appears the issue is on net-http2, using it alone fails sporadically with Socket was remotely closed (SocketError)
, retrying exactly the same request multiple times until the socket is not closed worked for me. Once the first request is "accepted" all the others never fail on that active connection. It can be the initial negotiation between the servers?
Hi. Is there a possibility that this case will be fixed with net-http2 timeout support? https://github.com/ostinelli/apnotic/commit/e30528bdd7194a183a9bbf235e98a3410f43bf80
This issue is around header ordering, which has been fixed. Any other issues with the same exception name should be re-opened as independent issues. It's also important to use .on(:error)
with the connection pool so closed connection can be caught and retried.
In the last 2 days, our production server has been raising "SocketError: Socket was remotely closed" whenever it tries to push to APN server. I didn't update production source code for weeks. It suddenly happens on all environments: production, staging, development. I also checked all Apple certificates. They are valid (not expired).
I am not alone. There are 2 people reporting about this issue on Apple forums: https://forums.developer.apple.com/thread/106078
I really appreciate if you know any thing about this issue.