CurrentContainerID is blank when /var/lib/docker/containers is a separate mount point on the host

[eigood]

docker-gen/internal/context/context.go

Line 231 in cfd2934

regex = fmt.Sprintf("containers/%v", regex)

My machine is currently not fetching .Docker.CurrentContainerID, it ends up being blank. I have a separate mountpoint for /var/lib/docker/containers, so any sub-relative paths don't have the /containers/ prefix in the container view.

2939 2016 0:96 / /dev/shm rw,nosuid,nodev,noexec,relatime - tmpfs shm rw,size=65536k,inode64
2940 2014 0:23 /docker.sock /tmp/docker.sock ro,relatime - tmpfs tmpfs rw,size=6578076k,mode=755,inode64
2941 2014 253:5 /17dc78431e3a010258883e916885bde27da979734063d00a5900a3ddd3ca61d8/hosts /etc/hosts rw,relatime - ext4 /dev/mapper/tooz--fast-docker--containers rw,errors=remount-ro,stripe=256
2942 2014 253:5 /17dc78431e3a010258883e916885bde27da979734063d00a5900a3ddd3ca61d8/resolv.conf /etc/resolv.conf rw,relatime - ext4 /dev/mapper/tooz--fast-docker--containers rw,errors=remount-ro,stripe=256
2943 2014 253:5 /17dc78431e3a010258883e916885bde27da979734063d00a5900a3ddd3ca61d8/hostname /etc/hostname rw,relatime - ext4 /dev/mapper/tooz--fast-docker--containers rw,errors=remount-ro,stripe=256
2944 2014 253:7 /nginx_nginxproxy-conf/_data /etc/nginx/conf.d rw,relatime master:62 - ext4 /dev/mapper/tooz--fast-docker--volumes rw,errors=remount-ro,stripe=256
2945 2014 253:7 /nginx_nginxproxy-vhost/_data /etc/nginx/vhost.d ro,relatime master:62 - ext4 /dev/mapper/tooz--fast-docker--volumes rw,errors=remount-ro,stripe=256

[eigood]

I can confirm that if I don't put a separate LVM on /var/lib/docker/containers, then docker-gen works and finds it's own container id.

[Hulkstern]

I also stumbled upon this issue inadvertently ran into when I setup zfs for my /var/lib/docker directory as well as a dataset for every sub directory of /var/lib/docker

To help those who are also having the same issue I was. Since I setup a zfs dataset for every sub directory of my /var/lib/docker my jwilder/nginx-proxy container was no longer able to generate the /etc/nginx/conf.d/default.conf file correctly since docker-gen could not correctly refer to containers. Which then broke my reverse proxy for my services/sites. I would simply get a 502 error with the nginx log entry of

no live upstreams while connecting to upstream, client: [CLIENT-IP], server: [MY-DOMAIN], request: "GET / HTTP/1.1", upstream: "http://[MY-DOMAIN]/", host: "[MY-DOMAIN]", referrer: "http://[MY-DOMAIN]/"

Hopefully the smattering of keywords I've put here will help someone googling a solution to the problem this behavior causes for the reverse-proxy image.