systemd: nginx does not resolve /etc/hosts entries at boot time

Question

Using Debian 9, the behaviour of systemd is the following:

• At boot-time nginx does not seems to read /etc/hosts and failed to start
• After boot, when the command systemctl start nginx it is working properly

Note: the file /etc/hosts is correct and remains untouched during all the tests.

After searching for a while, I tried to start systemd-resolved before nginx. There is no impact.

Oct 13 11:14:01 server-dev systemd-resolved[466]: Positive Trust Anchors:
Oct 13 11:14:01 server-dev systemd-resolved[466]: . IN DS 19036 8 2 49aac11d7b6f6446702e54a1607371607a1a41855200fd2ce1cdde32f24e8fb5
Oct 13 11:14:01 server-dev systemd-resolved[466]: . IN DS 20326 8 2 e06d44b80b8f1d39a95c0b0d7c65d08458e880409bbc683457104237c7f8ec8d
Oct 13 11:14:01 server-dev systemd-resolved[466]: Negative trust anchors: 10.in-addr.arpa 16.172.in-addr.arpa 17.172.in-addr.arpa 18.172.in-addr.arpa 19.172.in-addr.arpa 20.172.in-addr.arpa 21.172.in-addr.arpa 22.172.in-addr.arpa 23.172
Oct 13 11:14:01 server-dev systemd-resolved[466]: Using system hostname 'server-dev'.
Oct 13 11:14:01 server-dev systemd-resolved[466]: Switching to system DNS server 192.168.1.1.
Oct 13 11:14:01 server-dev shorewall[465]: Compiling using Shorewall 5.0.15.6...
Oct 13 11:14:01 server-dev systemd[1]: Started Network Name Resolution.
Oct 13 11:14:01 server-dev systemd[1]: Starting A high performance web server and a reverse proxy server...
Oct 13 11:14:01 server-dev systemd[1]: Reached target Multi-User System.
Oct 13 11:14:01 server-dev systemd[1]: Reached target Graphical Interface.
Oct 13 11:14:01 server-dev systemd[1]: Starting Update UTMP about System Runlevel Changes...
Oct 13 11:14:01 server-dev systemd[1]: Started Update UTMP about System Runlevel Changes.
Oct 13 11:14:01 server-dev nginx[502]: nginx: [emerg] host not found in upstream "server-dev.com" in /etc/nginx/sites-enabled/default:33
Oct 13 11:14:01 server-dev nginx[502]: nginx: configuration file /etc/nginx/nginx.conf test failed
Oct 13 11:14:01 server-dev systemd[1]: nginx.service: Control process exited, code=exited status=1
Oct 13 11:14:01 server-dev systemd[1]: Failed to start A high performance web server and a reverse proxy server.
Oct 13 11:14:01 server-dev systemd[1]: nginx.service: Unit entered failed state.
Oct 13 11:14:01 server-dev systemd[1]: nginx.service: Failed with result 'exit-code'.

Any suggestion?

Answer 1

I found a working solution, it is linked to nginx proxy configuration. I am not clear about the impacts, feel free to comment or propose a better solution.

1. Install dnsmasq

2. Add a domain name resolver in nginx configuration

server {
# redirect DNS requests to local dnsmasq
resolver 127.0.0.1;

# without these 2 lines, nginx does not resolve name correctly
set $mt_proxy_from "http://server-dev.com:8000";
set $mt_proxy_dest "https://server-dev.com";
[...]
location / {
# Proxy configuration
proxy_pass $mt_proxy_from;
proxy_read_timeout 90;
proxy_redirect $mt_proxy_from $mt_proxy_dest;
}}

I don't like this solution because it seems not efficient:

• Install dnsmasq and add a resolver to nginx sounds OK
• Add NGINX dynamic variables to resolve the names dynamically (not clear when?) seems weird because the local addresses are fixed

Question: is there any way to do something clean?

• Configure the IP addresses in /etc/hosts
• Make NGINX resolve the names according /etc/hosts at systemd boot time?

Sources:
https://stackoverflow.com/questions/8305015/when-using-proxy-pass-can-etc-hosts-be-used-to-resolve-domain-names-instead-of
https://www.jethrocarr.com/2013/11/02/nginx-reverse-proxies-and-dns-resolution/