https://bz.apache.org/bugzilla/show_bug.cgi?id=63170
--- Comment #22 from Warren D. Johnson <war...@jtechgrp.com> ---
Sorry about all the comments here but I did a little deeper inspection of the
situation. I tested my http2 setup on the website's server by browsing
directly to the website server (in other words, I bypassed the reverse proxy).
I notice something different in the trace8 logging. I am seeing this message
[Mon Feb 25 00:07:44 2019] [debug] [pid 31021] ssl_engine_io.c(1103):
[client 172.31.13.84:60614] AH02001: Connection closed to child 66 with
standard shutdown (server www.example.com:443)
I do NOT see that message when browsing the site through the reverse proxy.
This message is generated in ssl_engine_io.c line 1102. The function is
ssl_filter_io_shutdown(...). I am lead to believe one of two things is
happening:
1) ssl_filter_io_shutdown is not being called when browsing website through
reverse proxy.
or
2) We are abnormally returning from the function through the following clause
situated in the beginning of the function
if (!ssl) {
return;
}
ssl refers to
SSL *ssl = filter_ctx->pssl;
from
ssl_filter_io_shutdown(ssl_filter_ctx_t *filter_ctx,
conn_rec *c, int abortive)
In the case of when I browse directly to the client website (no reverse proxy),
the process appears to be "Apache and h2 service the request.... all requests
done, request is closed ..... wait for KeepAliveTimeout, then h2 terminates the
connection".
In the case of when I browse the client website through the reverse proxy, the
process appears to be "Apache and h2 service the requests". There is no
acknowledgement of "Connection Closed". Instead we ONLY wait for
KeepAliveTimeout and then close connection.
Meanwhile, the Reverse proxy is also waiting for the KeepAliveTimeout (as it
relates I think to requests from the person doing the web browsing) and then
terminates. h2 on client website seems to be fine waiting, skipping
"Connection Closed" and then timing out. On the other hand, the reverse proxy
h2 seems to freak out and use a lot of CPU in the process while it's
waiting....... maybe some kind of crazy looping going on somewhere? I can't
really see anything in the logs about it.
Thanks!
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org
