Error response from daemon: endpoint with name XXX already exists in network bridge

I think the #1642 (closed) bug is back in gitlab-ci-multi-runner 9.1.0 or somehow this is related to that. (Previously we had that bug as well.)

Summarize the bug encountered concisely

After one of the pipeline dies, all the next ones are failing.

How one can reproduce the issue - this is very important

After a while one of the pipelines died right after finishing successfully the test:

...
ERROR: Job failed (system failure): Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?

After then every pipeline dies:

Running with gitlab-ci-multi-runner 9.1.0 (0118d89)
  on xxxx (xxx)
Using Docker executor with image xxxx ...
Using docker image xxxx for predefined container...
WARNING: Possibly zombie container runner-196bff67-project-127-concurrent-1-predefined is disconnected from network bridge
Pulling docker image xxxx ...
Using docker image xxxx ID=xxx for build container...
ERROR: Job failed (system failure): Error response from daemon: endpoint with name runner-196bff67-project-127-concurrent-1-predefined already exists in network bridge

or

ERROR: Job failed (system failure): Error response from daemon: endpoint with name runner-196bff67-project-127-concurrent-0-predefined already exists in network bridge

There is no running pipeline:

root@xxxx # docker ps
CONTAINER ID        IMAGE                COMMAND                  CREATED             STATUS              PORTS                    NAMES
713b60852b1b        minio/minio:latest   "go-wrapper run se..."   6 months ago        Up About an hour    0.0.0.0:9005->9000/tcp   minio

I cannot list that network bridge:

root@xxxx # docker network ls
NETWORK ID          NAME                DRIVER              SCOPE
2a9a9f88f8b7        bridge              bridge              local
6b9f08506376        host                host                local
9b4f1ecf5c1c        none                null                local

The same happens after docker network prune

Environment description

docker info

Containers: 21
 Running: 1
 Paused: 0
 Stopped: 20
Images: 16
Server Version: 17.04.0-ce
Storage Driver: aufs
 Root Dir: /var/lib/docker/aufs
 Backing Filesystem: extfs
 Dirs: 277
 Dirperm1 Supported: false
Logging Driver: json-file
Cgroup Driver: cgroupfs
Plugins: 
 Volume: local
 Network: bridge host macvlan null overlay
Swarm: inactive
Runtimes: runc
Default Runtime: runc
Init Binary: 
containerd version: 422e31ce907fd9c3833a38d7b8fdd023e5a76e73
runc version: 9c2d8d184e5da67c95d601382adf14862e4f2228
init version: 949e6fa
Security Options:
 apparmor
 seccomp
  Profile: default
Kernel Version: 3.12.0-7-generic
Operating System: Ubuntu 16.04.2 LTS
OSType: linux
Architecture: x86_64
CPUs: 2
Total Memory: 7.552GiB
Name: nci
ID: DIDT:JASO:SC6M:I4EP:YOMT:2WMA:GJFS:ZOMT:K4NA:QPGR:PFC5:3MRB
Docker Root Dir: /var/lib/docker
Debug Mode (client): false
Debug Mode (server): false
Registry: https://index.docker.io/v1/
Experimental: false
Insecure Registries:
 127.0.0.0/8
Live Restore Enabled: false

Used GitLab Runner version

Version:      9.1.0
Git revision: 0118d89
Git branch:   9-1-stable
GO version:   go1.7.5
Built:        Fri, 21 Apr 2017 14:32:04 +0000
OS/Arch:      linux/amd64