`Ta` increase after 2.3 -> 2.4 upgrade

Help!
1

Hi, I hope someone can help me out with this question. Recently we’ve upgraded from 2.3 to 2.4. We log and monitor using the default timing events TR/Tw/Tc/Tr/Ta. After upgrading to 2.4, the TR (and of course Ta) have increased dramatically to up to 30 seconds, but only for HTTP/2 connections. I don’t find anything in the documentation of the timing events that would explain these long connections. What would be the explanation for these high TR values and what would be a way to deal with it?

Our configuration looks something like this:

# cat /etc/haproxy/haproxy.cfg 
# Ansible managed
global
  ...
defaults
  ...
listen stats
  ...
listen ssl-proxy
  timeout http-keep-alive 5s
  server socket-ssl-proxy /run/haproxy/ssl-proxy.sock send-proxy-v2
listen ssl-proxy-client-cert-auth
  timeout http-keep-alive 5s
  server socket-ssl-proxy /run/haproxy/ssl-proxy.sock send-proxy-v2
frontend http
  bind /run/haproxy/ssl-proxy.sock mode 0660 accept-proxy
  timeout http-keep-alive 5s
backend php-servers
  option http-server-close
backend cdn-servers
  option http-server-close

Our backends contain Apache and Nginx servers with keepalive enabled.

Apache configuration looks something like this:

#
# KeepAlive: Whether or not to allow persistent connections (more than
# one request per connection). Set to "Off" to deactivate.
#
KeepAlive On

#
# MaxKeepAliveRequests: The maximum number of requests to allow
# during a persistent connection. Set to 0 to allow an unlimited amount.
# We recommend you leave this number high, for maximum performance.
#
MaxKeepAliveRequests 100

#
# KeepAliveTimeout: Number of seconds to wait for the next request from the
# same client on the same connection.
#
KeepAliveTimeout 2

Nginx uses the default keepalive settings.