Since yesterday I have been experiencing random failures of nginx during reload, which leave this in the journal:
Nov 21 10:26:16 nixos systemd[1]: Reloading Nginx Web Server.
Nov 21 10:26:16 nixos systemd[4982]: Mount point source '/tmp/systemd-private-3895d60d10d54499bd64379bfe092b83-nginx.service-TgjKSh/tmp' is not accessible: No such file or directory
Nov 21 10:26:16 nixos systemd[4982]: Failed to mount /tmp/systemd-private-3895d60d10d54499bd64379bfe092b83-nginx.service-TgjKSh/tmp to /run/systemd/unit-root/tmp: No such file or directory
Nov 21 10:26:16 nixos systemd[4982]: nginx.service: Failed to set up mount namespacing: /run/systemd/unit-root/tmp: No such file or directory
Nov 21 10:26:16 nixos systemd[4982]: nginx.service: Failed at step NAMESPACE spawning /nix/store/5kla5jnb0q2jbmzv61dr0i21cgmxa2dx-nginx-1.18.0/bin/nginx: No such file or directory
Nov 21 10:26:16 nixos systemd[1]: nginx.service: Control process exited, code=exited, status=226/NAMESPACE
I have absolutely no idea where this comes from and so far I have been unable to construct a minimal reproducer. This is very annoying because usually I deploy my server with NixOps but when nginx fails during reload it takes all the ACME and anything that’s reverse-proxied with it, so I have to manually log in an restart a whole bunch of units by hand.
Is anyone else experiencing or has experienced something similar and can tell me what I need to fix?