Nginx reverse proxy losts connection to PHP container after upgrade

I have two containers inside a Rancher stack. One is php-fpm container which is serving an application. The second one is Nginx acting as a reverse proxy.

The nginx has following configurations mounted:

/etc/nginx/nginx.conf

user nginx;
worker_processes 4;
pid /var/run/nginx.pid;

events {
  worker_connections  2048;
}

http {
  server_tokens off;
  sendfile on;
  tcp_nopush on;
  tcp_nodelay on;
  keepalive_timeout 15;
  types_hash_max_size 2048;
  include /etc/nginx/mime.types;
  default_type application/octet-stream;
  include /etc/nginx/conf.d/*.conf;
  open_file_cache max=100;
}

/etc/nginx/conf.d/default.conf

server {
    ...
    location ~ ^/(app|app_dev|config)\.php(/|$) {
        fastcgi_param HTTP_PROXY "";
        fastcgi_pass php:9000;
        fastcgi_split_path_info ^(.+\.php)(/.*)$;
        include fastcgi_params;
        fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
        fastcgi_param HTTPS off;
    }
}

Everything works well until the php container is upgraded. From that moment on the reverse proxy returns 502 Bad Gateway. Following message is logged in the error log:

*5 connect() failed (113: Host is unreachable) while connecting to upstream,
client: 10.42.154.177, server: [hidden url],
request: "GET / HTTP/1.1", upstream: "fastcgi://10.42.241.63:9000",
host: "[hidden-url]"

So instead of using the hostname, nginx uses IP directly which obviously gets changed during container upgrade. How to make this work, always?

I could create health checks so nginx container would get recreated however my error logs get flooded with messages.