Open con-f-use opened 5 years ago
I am seeing a similar or the same issue - log full of messages like this:
[uwsgi-http key: client_addr: 10.0.0.14 client_port: 4797] hr_ssl_read(): Connection reset by peer [plugins/http/https.c line 393]
with various ephemeral ports for the client_port
value
In my case the client_addr
is an AWS Network Load Balancer - so I am assuming it is connecting on various ephemeral ports using HTTP to my uwsgi instance, which is configured only for HTTPS as follows:
buffer-size = 32768
chdir = /project/src
disable-logging = 0
enable-threads = 1
gid = www-data
https = =0,/project/ssl/fullchain.pem,/project/ssl/privkey.pem,HIGH
master = 1
module = remetrik.wsgi
pcre-jit = 1
post-buffering = 1
post-buffering-bufsize = 32768
processes = 1
shared-socket = 0.0.0.0:443
ssl-verbose = 1
stats = :1717
stats-http = 1
strict = 1
threads = 1
thunder-lock = 1
uid = www-data
Environment is latest django+uwsgi (no nginx) running in an alpine docker container on AWS Fargate/ECS.
I'm unsure yet whether this is causing a problem or not beyond the log messages (currently debugging other problems).
Hi @tunecrew have you solved this problem? I am seeing the same issue
[uwsgi-http key: client_addr: 192.241.207.93 client_port: 5813] hr_ssl_read(): Connection reset by peer [plugins/http/https.c line 393]
And i am using Azure k8s to deploy my web application, here is my uwsgi(no nginx) configuration:
[uwsgi]
master = true
https = 0.0.0.0:5001,/root/certs/tls.crt,/root/certs/tls.key,HIGH
module = https_manage:app
pidfile = /app/uwsgi.pid
processes = 4
threads = 2
vacuum = true
http-timeout = 300
Hi @Bob-CN we ended up moving to Daphne so we could use asgi, so no.
@tunecrew Got it, thanks!
Currently, uwsgi gives a
Connection reset by peer
response, when you try to make a HTTP-request on an HTTPS port:That is true even when re-rourting:
This should be handled more gracefully.