Docker Community Forums

Share and learn in the Docker community.

Error starting userland proxy + container unable to launch at reboot ubuntu

Hi,

One of my usage of docker is through a ubuntu host.
I don’ tthink the version is really a problem here.
The problem is that the container which are supposed to be restarted at boot and which have specific IP to answer to like -p 10.0.0.10:10000 are not doing it because of this (this is jsut an example as many others)

Aug 14 00:14:27 kubuntu dockerd[1722]: time="2019-08-14T00:14:27.214758962Z" level=error msg="Failed to start container 6a26e5ddccd0da424cd08998800c0f7630b2584604bb70998154068c5b1090eb: driver failed programming external connectivity on endpoint plex (fc53161481008076225365328e621af5ca74c6a6678a30c5928925cd28387cd2): Error starting userland proxy: listen tcp 10.0.0.100:32469: bind: cannot assign requested address"
Aug 14 00:14:27 kubuntu dockerd[1722]: time="2019-08-14T00:14:27.242410709Z" level=warning msg="d5095674f2a7122080f4d49781d21bd85ccbd8fe523630775380b985471dee9c cleanup: failed to unmount IPC: umount /var/lib/docker/containers/d5095674f2a7122080f4d49781d21bd85ccbd8fe523630775380b985471dee9c/mounts/shm, flags: 0x2: no such file or directory"
Aug 14 00:14:27 kubuntu dockerd[1722]: time="2019-08-14T00:14:27.242984461Z" level=error msg="d5095674f2a7122080f4d49781d21bd85ccbd8fe523630775380b985471dee9c cleanup: failed to delete container from containerd: no such container"
Aug 14 00:14:27 kubuntu dockerd[1722]: time="2019-08-14T00:14:27.243004700Z" level=error msg="Failed to start container d5095674f2a7122080f4d49781d21bd85ccbd8fe523630775380b985471dee9c: driver failed programming external connectivity on endpoint Sync (9c94d1d7fcccfee21ab7666c843ab47cc71d17c4dcee04665ba6cb17a73dcc1e): Error starting userland proxy: listen tcp 10.0.0.100:50100: bind: cannot assign requested address"

It has been I think the same error since 16.04 at least. And if my experiences are correct it doesn’t happen if I don’t put a specific ip to those container.
It’s not a problem of the ip or port being in used. It’s a problem of timer. The docker engine is loaded too early in the boot process. The best proof of it is that if I restart the docker.service then every container start correctly.
I don’t see anyone else with the same problem bbut it seems quite a common problem to me, so is there a trick or a workaround to resolve this?
I don’t use docker-compose, I’m just launching my container through command lines.
Should I file an issue for that on github? I’ve seen a lot of people having this problem but with windows version.

thanks in advance for all the help and avices

still no one has any answer about that?
Is it specific to ubuntu? I4ve seen on github that a lot of windows user have the same problem.
Does anyone have a logic explanation why the docker engine is trying to restart a container and so a connection before the whole networking setup from netplan.io is in place?
I would really appreciate an answer from someone on the devs on that?