On one of my Docker servers that is physical, I have an issue by reboot: some containers that are meant to restart automatically with restart “always” are not.
By looking at logs of docker service, I have the following message for all 3 of them:
This wasn’t always like that, it was working a couple of months ago so I don’t know what changed except some updates. Same configuration/container/app with Docker on an Almalinux VM doesn’t have this issue at boot (I have a couple of them, none is impacted)
Almalinux version: AlmaLinux 9.5 (fully patched as today)
Hardware: Minisforum MS01 / Intel Corporation Ethernet Controller X710 (integrated NICs)
Network config: LACP trunk with several vlans
All containers are using macvlan IPs on differents VLANs/subnets
Containers starts without any issue right after boot, but I have to start them manually.
The above config postpones the start of ‘kadmin.service’ until after ‘dirsrv@ldap1.service’.
NetworkManager-wait-online.service might be the service that – when “started” – means that all normal network bits are ok. (If is Docker already set to wait for that or something similar, then I have no idea what to test next.)
Thanks for you answer. At some point, this is also what I’m thinking, but I don’t know which service(s) I should add to the dependency list. Adding NetworkManager-wait-online.service didn’t solve the issue unfortunatelly.
Here the current list of docker dependencies including the one just added):
PS: restarting the docker service start all the “always” restarting docker containers.
Also, if I don’t stop any of the containers (there is around 50 on this host), a part of them will be started, so it confirms somehow that “something” needs to be ready for docker, but is not when docker service boots