Loki driver prevent some containers to start on reboot

Grafana Loki
1

Hello,

Since I install loki driver in docker rootless daemon, some containers not restart on reboot. I have the error:

dockerd-rootless.sh[543]: time="2023-07-11T23:57:16.549694424+02:00" level=error msg="5ef8f956332b0e5259fdf500fa65c957d21692518b671bea0de982cf45ffa58c cleanup: failed to delete container from containerd: container \"5ef8f956332b0e5259fdf500fa65c957d21692518b671bea0de982cf45ffa58c\" in namespace \"moby\": not found"

I must restart docker to resolv the problem.

Why loki prevent some containers to start?
What can be done to circumvent the problem?

Thank you

2

I found the culprit.
I use systemd services to start some containers of monitoring like node exporter, cadvisor, etc. This blocks other containers that were in exited status at startup. I thought it was the loki driver because it was during installation that the problem appeared.
I don’t know why it conflicts with systemd, but at least it’ll be easier to investigate.
Example:

cat .config/systemd/user/docker.node.service
[Unit]
Description=Node exporter Service
After=docker.service
Requires=docker.service

[Service]
TimeoutStartSec=0
Restart=always
ExecStartPre=-docker stop node
ExecStartPre=-docker rm node
ExecStartPre=docker pull prom/node-exporter:latest
ExecStart=docker run --rm --name node -p 9100:9100 prom/node-exporter

[Install]
WantedBy=default.target