/etc/letsencrypt/live/ empty, nginx won't start.

ZinkDifferent

suddenly, all certificates for all sites on my server have disappeared. The location

/etc/letsencrypt/live/

is empty. As a result of this situation, nginx won't start since it keeps encountering errors of missing certificates. I've attempted to reissue certificates, but since acme can't verify webfoot since nginx isn't running, I can't do that.

HELP!

What's the best way to get around this? Either by forcing certificates to be issued even with the server not running (--force doesn't do it), or by 'en masse' removing all the certificates from all hosted domains, so nginx can start again, and subsequently issuing all certificates 'en masse' again.

ZinkDifferent

No sure what led to this - I've only done a purge of php80m, php81 and Redis (since I had that Redis error that could only be fixed that way), then performed a regular apt dist-update, and a reboot. None of which should have messed with certificates.

Please advise.

ZinkDifferent

Finally, I have disabled all sites on that server, and nginx still won't start.

I now get:

nginx: [emerg] no port in upstream "multiphp" in /etc/nginx/sites-enabled/22222:47

and that line is:

  location ~ \.php$ {
    try_files $uri =404;
    include fastcgi_params;
    fastcgi_pass multiphp;
  }

At this point, I've exhausted my possible fixes. I've even reinstalled words, but that did nothing. I don't want to purge nginx, as I don't want to lose the site configurations, obviously.

marty

multiphp is defined in /etc/nginx/conf.d/upstream.conf - what do you have in there?
It should look like e.g.

upstream multiphp {
    server unix:/var/run/php/php73-fpm.sock;
    server unix:/var/run/php/php74-fpm.sock;
    server unix:/var/run/php/php72-fpm.sock;
    server unix:/var/run/php/php80-fpm.sock;
    server unix:/var/run/php/php81-fpm.sock;
}