Proxy host nginx configs not regenerated on container restart

[jordanlambrecht]

After a routine container restart, all proxy host nginx configuration files in /data/nginx/proxy_host/ are missing, even though the proxy hosts still appear in the web UI. This causes all proxied domains to fail with SSL errors (SSL_ERROR_UNRECOGNIZED_NAME_ALERT).

At first I was looking at #2549 but I think this is a distinct issue. that issue involved a crash loop after an update with directive errors. In my case, there are no errors, nginx passes config test, but the proxy_host/ directory is simply empty. The application starts cleanly but never generates the configs from the database.

The /data/nginx/proxy_host/ directory is empty after restart:

$ docker exec proxymanager-plus ls -la /data/nginx/proxy_host/
total 0
drwxr-xr-x 1 root root  64 Mar 13 10:13 .
drwxr-xr-x 1 root root 320 Jan 19 14:18 ..

The database still contains all hosts:

$ docker exec proxymanager-plus ls -la /data/npmplus/
total 480
drwxrwx--- 1 root root    256 Mar 13 13:51 .
drwxr-xr-x 1 root root    352 Jan 19 12:19 ..
-rwxrwx--- 1 root root 372736 Mar 13 13:51 database.sqlite
...

Hosts appear correctly in the web UI. Clicking "Save" on any individual host regenerates its config file, confirming the data is intact.

My docker compose:

proxymanager-plus:
  image: docker.io/zoeyvid/npmplus:latest
  container_name: proxymanager-plus
  hostname: proxymanager-plus
  network_mode: host
  restart: always
  cap_add:
    - NET_ADMIN
    - NET_BIND_SERVICE
    - SYS_MODULE
    - BPF
    - PERFMON
  volumes:
    - "./npmplus/opt/npmplus:/data"
  environment:
    TZ: America/Chicago
    ACME_EMAIL: [redacted]
    DISABLE_IPV6: true
    NGINX_LOG_NOT_FOUND: true
    LOGROTATE: true
    LOGROTATIONS: 7
    GOA: true
    NGINX_LOAD_GEOIP2_MODULE: true

No errors in logs during startup. Nginx config test passes:

$ docker exec proxymanager-plus nginx -t
nginx: the configuration file /usr/local/nginx/conf/nginx.conf syntax is ok
nginx: configuration file /usr/local/nginx/conf/nginx.conf test is successful

[Zoey2936]

you can one time set REGENERATE_ALL=true as env at startup, this will regenerate all config files, but only set it once and remove it again, since otherwise your configs will be recreated on any container restart. Maybe this happended because the container startup was cancled before it could generate the configs?

[jordanlambrecht]

Actually just did some more testing and wanted to confirm for the record that it's not happening every restart or up/down, so you might be right. That wouldn't explain why it wouldn't self-heal and rebuild the configs after a fresh docker compose down && docker compose up -d though.

This is, however, the second time this has happened. Happened a few weeks ago out of the blue after seemingly no intervention on my part.