• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

Issue with Docker service after last update

Hi,

The output was too long and I couldnt paste it here. I have uploaded output to txt in gdrive link below.
plesk-docker.txt
I checked the logs, it looks like a Docker service can be started if you executed it from CLI, I had the same logs on my server. Something wrong if you launched it as a systemd service.

@AYamshanov my license is from ovh and I think support is not included.
This seems to be bug related to last update. How should I proceed?
How to get technical support for Plesk

---

Could someone provide steps-to-reproduce the issue? Last week, I installed the Plesk 17.8 with the previous version of Docker extension (1.3.2-126), deployed a container from Docker Hub, upgrade the extension to the latest (1.4.0-138), restarted containers, restarted my server. After these steps, I checked services, the container - all worked.

Thanks @Erwan for Forwarded to devs - Docker service stopped after last Plesk update .
 
AYamshanov ,

Saturday, we have made the plesk update on 20 servers with 3 servers with Plesk Docker services.
The new version is: Plesk Onyx 17.8.11 #22.
We don't know what was the previous version. In fact, the stable version at this date: 2018-09-16 (last Plesk update - we made it always manually).

After this update, we have made a kernel update:
#yum update

That's all.
The servers are all under Centos 7.3.1611.

And the Plesk Docker service was down. We we load the "Docker" menu:
Error: Docker service is not available: No such file or directory

And since, all the services was down...

Do you really have no idea / solution?
These services are important for our customers...
 
After that issue I have decided to run plesk under virtualization. With daily backups would easier to manage.
 
@bulent,

Running Plesk under virtualization doesn't change anything for this problem.
On our side, we use Plesk Docker to make use ownCloud.

@Yamshanov:
It's absolutely necessary that we find a solution, our customers can not wait.
It can be a solution to remove /var/lib/docker? Job for docker.service failed because the control process exited with error code · Issue #162 · docker/for-linux

Process?
1) copy "OwnCloud data" => /var/lib/docker/volumes/ed623eb3ede5833160456cccfb523a5ca58af28d70cf8707f3103894da6a111a/_data/
to /var/tmp (to restore the files next)
2) delete /var/lib/docker: #rm -rf /var/lib/docker
3) restart Docker
 
@Erwan the idea of virtualization is to have daily snapshots and after such issue, just restore the last one and disable the updates before new update comes out.
 
So we have resolved the problem (temporarily) with a kernel rollback.

# yum history
Modules complémentaires chargés : fastestmirror
ID | Identifiant utilisateur | Date et heure | Action | Modifié
-------------------------------------------------------------------------------
41 | root <root> | 2018-11-10 14:14 | I, U | 19
40 | Système <indéfini> | 2018-11-10 14:10 | Update | 78
39 | Système <indéfini> | 2018-11-10 14:10 | Update | 1
38 | root <root> | 2018-10-19 03:32 | I, O | 4 EE

# yum history rollback 39

And i now no problem to load the service...

it seems that the problem is a depreciation of "Legacy “overlay” storage driver" or "device mapper storage driver" with the Docker v18.09.0 .
 
The root cause; few different actions in one place, at the same time
  • If you run Docker service in another container like LXC (nested containers), you can't use "overlay" fs kernel module. I am not an expert in LXC, but as I know, you need to do something to resolve this case. So, by default, you can't use overlay; maybe because of security but I am not sure.
  • In the latest update of the extension, old abandoned docker repository was switched to actual Docker Engine (version 18); Now if something new released, it can be installed on your system by packet manager like yum or apt; your docker service is up-to-date.
  • Five days ago, docker-ce v18.09 was released
Changelog:
[...]
- Packaging
- - Split "engine", "cli", and "containerd" to separate packages, and run containerd as a separate systemd service docker-ce-packaging#131, docker-ce-packaging#158
[...]


What's happened:
  1. Plesk updated the docker extension, the Docker extension updated repository, a packet manager updated docker to version 18 (in October, it was docker-ce 18.06)
  2. When docker-ce 18.09 was released, the packet manager updated docker to 18.09.
  3. Docker 18.09 contains a new separate file for containerd.service. This service pre-check overlay kernel module in a system; if the module doesn't exist, the service doesn't start. In LXC, it doesn't exist.
  4. If containerd doesn't start, docker service doesn't start too because of 'dependency'.

Workaround:
As @Erwan said, rollback to the previous version of docker package resolves the problem.


 
Hello,
I have exact the same issue (I'm from OVH too). I successfully made the rollback but I can't install the older version of the extension:

upload_2018-11-14_2-28-59.png

After I press Install:
upload_2018-11-14_2-30-3.png

In Docker tab: "This KB article" redirects to -> Plesk Help Center
upload_2018-11-14_2-30-26.png
 
Back
Top