After disconnecting and reconnecting a container, the exposed port does not rebind to the new host port
Description After disconnecting and reconnecting a container to the bridge network with a dynamic host port binding, the newly assigned host port is not reachable and the container is still reachable via the old port binding.
Steps to reproduce the issue:
- Run
docker run --name mynginx1 -p 80 -d nginx - Inspect the container and telnet the binded port
- Run
docker network disconnect bridge <container Id> - Run
docker network connect bridge <container Id>
Describe the results you received:
a. Try to telnet container on newly binded host port (Identify it is unreachable) b. Try to telnet container on old binded host port (Identify it is reachable)
Describe the results you expected: a. Try to telnet container on newly binded host port (Identify it is reachable) b. Try to telnet container on old binded host port (Identify it is unreachable)
Additional information you deem important (e.g. issue happens only occasionally):
Output of docker version:
Client: Docker Engine - Community
Cloud integration 0.1.18
Version: 19.03.13
API version: 1.40
Go version: go1.13.15
Git commit: 4484c46d9d
Built: Wed Sep 16 17:00:27 2020
OS/Arch: windows/amd64
Experimental: false
Server: Docker Engine - Community
Engine:
Version: 19.03.13
API version: 1.40 (minimum version 1.12)
Go version: go1.13.15
Git commit: 4484c46d9d
Built: Wed Sep 16 17:07:04 2020
OS/Arch: linux/amd64
Experimental: false
containerd:
Version: v1.3.7
GitCommit: 8fba4e9a7d01810a393d5d25a3621dc101981175
runc:
Version: 1.0.0-rc10
GitCommit: dc9208a3303feef5b3839f4323d9beb36df0a9dd
docker-init:
Version: 0.18.0
GitCommit: fec3683
Output of docker info:
Client:
Debug Mode: false
Server:
Containers: 10
Running: 4
Paused: 0
Stopped: 6
Images: 1755
Server Version: 19.03.13
Storage Driver: overlay2
Backing Filesystem: extfs
Supports d_type: true
Native Overlay Diff: true
Logging Driver: json-file
Cgroup Driver: cgroupfs
Plugins:
Volume: local
Network: bridge host ipvlan macvlan null overlay
Log: awslogs fluentd gcplogs gelf journald json-file local logentries splunk syslog
Swarm: inactive
Runtimes: runc
Default Runtime: runc
Init Binary: docker-init
containerd version: 8fba4e9a7d01810a393d5d25a3621dc101981175
runc version: dc9208a3303feef5b3839f4323d9beb36df0a9dd
init version: fec3683
Security Options:
seccomp
Profile: default
Kernel Version: 4.19.104-microsoft-standard
Operating System: Docker Desktop
OSType: linux
Architecture: x86_64
CPUs: 12
Total Memory: 24.95GiB
Name: docker-desktop
ID: TAE7:3G7R:YSMF:AJDH:OPMK:CFTK:74OY:LJ3M:OFOH:KI4S:TL4I:CXO7
Docker Root Dir: /var/lib/docker
Debug Mode: true
File Descriptors: 74
Goroutines: 71
System Time: 2021-01-04T20:38:06.7034025Z
EventsListeners: 2
Registry: https://index.docker.io/v1/
Labels:
Experimental: false
Insecure Registries:
127.0.0.0/8
Live Restore Enabled: false
Product License: Community Engine
WARNING: bridge-nf-call-iptables is disabled
WARNING: bridge-nf-call-ip6tables is disabled
Additional environment details (AWS, VirtualBox, physical, etc.): We are having the same behavior across different types of host OS
going through tickets, and I can reproduce this;
- Looks to be only with randomly (ephemeral) ports
- :information_source: Situation is slightly different on Docker Desktop, likely because of extra proxies in between (host <--> VM)
- :question: not sure what the expected behavior is on this one; it may be because the ephemeral port is re-selected on connect?
- :warning: but on Desktop, there may be a bug because the port doesn't seem to be released (while disconnected)
1. Docker Desktop 4.41.0 (190386), manually defined port-mapping;
docker run --name mynginx1 -p 8080:80 -d nginx:alpine
9e072c150b63983f90a3e352fa1bf517dcd4db1dcb9a67063f7a5ec325b0a511
curl -I localhost:8080
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 21:42:00 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
docker network disconnect bridge mynginx1
curl -I localhost:8080
curl: (56) Recv failure: Connection reset by peer
docker network connect bridge mynginx1
curl -I localhost:8080
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 21:42:48 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
docker rm -fv mynginx1
curl -I localhost:8080
curl: (7) Failed to connect to localhost port 8080 after 0 ms: Couldn't connect to server
2. Docker Desktop 4.41.0 (190386), randomly assigned (ephemeral) port;
docker run --name mynginx1 -p 80 -d nginx:alpine
docker port mynginx1 80
0.0.0.0:49511
curl -I localhost:49511
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 21:45:23 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
disconnecting; note that the port appears to still be mapped, but doesn't give a response (unlike for the manually mapped port, which gave curl: (56) Recv failure: Connection reset by peer);
`
docker network disconnect bridge mynginx1
curl -I localhost:49511
curl: (52) Empty reply from server
re-connecting to the network;
docker network connect bridge mynginx1
curl -I localhost:49511
curl: (52) Empty reply from server
removing the container removes the port;
docker rm -fv mynginx1
curl -I localhost:49511
curl: (7) Failed to connect to localhost port 49511 after 0 ms: Couldn't connect to server
3. On Ubuntu 24.10, manually defined port-mapping
docker run --name mynginx1 -p 8080:80 -d nginx:alpine
curl -I localhost:8080
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 21:59:55 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
disconnect:
docker network disconnect bridge mynginx1
curl -I localhost:8080
curl: (7) Failed to connect to localhost port 8080 after 0 ms: Couldn't connect to server
re-connect:
docker network connect bridge mynginx1
curl -I localhost:8080
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 22:00:50 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
remove:
docker rm -fv mynginx1
curl -I localhost:8080
curl: (7) Failed to connect to localhost port 8080 after 0 ms: Couldn't connect to server
4. On Ubuntu 24.10, random (ephemeral) port
docker run --name mynginx1 -p 80 -d nginx:alpine
docker port mynginx1 80
0.0.0.0:32768
[::]:32768
curl -I localhost:32768
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 21:52:55 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
disconnecting; note that the port is not mapped (failed to connect);
curl -I localhost:32768
curl: (7) Failed to connect to localhost port 32768 after 0 ms: Couldn't connect to server
reconnecting;
docker network connect bridge mynginx1
curl -I localhost:32768
curl: (7) Failed to connect to localhost port 32768 after 0 ms: Couldn't connect to server
removing;
docker rm -fv mynginx1
curl -I localhost:32768
curl: (7) Failed to connect to localhost port 32768 after 0 ms: Couldn't connect to server
❓ not sure what the expected behavior is on this one; it may be because the ephemeral port is re-selected on connect?
Confirming that that's indeed the cause, so I think this may be expected behavior; ephemeral ports are assigned when connecting to the network. When disconnecting, that port is released, and when re-attachiing to the network, a new port is selected;
First time, the port is 32770;
docker run --name mynginx1 -p 80 -d nginx:alpine
docker port mynginx1 80
0.0.0.0:32770
[::]:32770
curl -I localhost:32770
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 22:07:34 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
After disconnecting, no port is assigned;
docker network disconnect bridge mynginx1
docker port mynginx1 80
Error: No public port '80' published for mynginx1
Re-attaching to the network assigns a new port (32771);
docker network connect bridge mynginx1
docker port mynginx1 80
0.0.0.0:32771
[::]:32771
And connecting to the new port works;
curl -I localhost:32771
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 22:08:19 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
Docker Desktop (bug?)
On Docker Desktop, it looks like there's a bug; port-mapping isn't restored / gets out of sync;
First steps look correct;
docker run --name mynginx1 -p 80 -d nginx:alpine
docker port mynginx1 80
0.0.0.0:50356
curl -I localhost:50356
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 22:25:39 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
Inspecting the container also shows the port mapped;
docker inspect --format '{{json .NetworkSettings.Ports}}' mynginx1
{"80/tcp":[{"HostIp":"0.0.0.0","HostPort":"50356"}]}
And so does docker ps
docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
2b2eda5e33a8 nginx:alpine "/docker-entrypoint.…" 51 seconds ago Up 50 seconds 0.0.0.0:50356->80/tcp mynginx1
However, after disconnecting, Docker Desktop's port-mapping gets confused; the container no longer has a port mapped, but Docker Desktop is still listening on the port;
docker network disconnect bridge mynginx1
docker port mynginx1 80
Error: No public port '80' published for mynginx1
curl -I localhost:50356
curl: (52) Empty reply from server
Inspecting the container looks correct;
docker inspect --format '{{json .NetworkSettings.Ports}}' mynginx1
{}
When re-connecting to the network, things get more confusing; docker port doesn't show a port mapped, but also doesn't show an error ("No public port '80' published for mynginx1");
docker network connect bridge mynginx1
docker port mynginx1 80
# no response here
docker port mynginx1
# no response here
Inspecting the container show the port, but not mapped;
docker inspect --format '{{json .NetworkSettings.Ports}}' mynginx1
{"80/tcp":[]}
But docker ps DOES show a port;
docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
2b2eda5e33a8 nginx:alpine "/docker-entrypoint.…" 5 minutes ago Up 5 minutes 127.0.0.1:55008->80/tcp mynginx1
But it's not mapped;
curl -I localhost:55008
curl: (7) Failed to connect to localhost port 55008 after 0 ms: Couldn't connect to server
Trying a stop, start of the container looks to restore the expected behavior; a new port is assigned, and shows up in both docker ps, docker port, and docker inspect;
docker stop mynginx1
docker start mynginx1
docker port mynginx1 80
0.0.0.0:50431
docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
2b2eda5e33a8 nginx:alpine "/docker-entrypoint.…" 6 minutes ago Up 27 seconds 0.0.0.0:50431->80/tcp mynginx1
docker inspect --format '{{json .NetworkSettings.Ports}}' mynginx1
{"80/tcp":[{"HostIp":"0.0.0.0","HostPort":"50431"}]}
And after this, I can connect again;
curl -I localhost:50431
HTTP/1.1 200 OK
Server: nginx/1.27.5
Date: Thu, 24 Apr 2025 22:32:43 GMT
Content-Type: text/html
Content-Length: 615
Last-Modified: Wed, 16 Apr 2025 12:55:34 GMT
Connection: keep-alive
ETag: "67ffa8c6-267"
Accept-Ranges: bytes
@akerouanton @robmry PTAL 😄