-
Notifications
You must be signed in to change notification settings - Fork 5.8k
Description
Description
When Docker Engine uses the containerd image store (default for new installations since
Docker 28), docker compose build produces a new image ID on every rebuild even when no
files have changed. This causes docker compose up to unnecessarily recreate all containers
on every deploy.
Disabling the containerd image store on the same server fixes the issue — the second build
reuses the cache completely and the image ID stays the same.
Steps To Reproduce
Use Docker CE on Linux with the containerd image store enabled (default for fresh installs):
$ docker info -f '{{.Driver}}'
overlayfs
Create these three files:
Dockerfile:
FROM alpine:3.21
RUN echo "hello" > /hello.txt
COPY config.txt /config.txt
CMD ["cat", "/hello.txt", "/config.txt"]config.txt:
static config file
docker-compose.yml:
services:
app:
build: .
container_name: test-cacheThen run:
# Build twice with nothing changed in between:
docker compose build
docker images --format '{{.ID}}' --filter reference='*-app'
docker compose build
docker images --format '{{.ID}}' --filter reference='*-app'Expected behavior
Second build reuses cache. Image ID is the same. docker compose up does not recreate the
container.
Actual behavior
Second build produces a different image ID. docker compose up sees a different image and
recreates the container.
Note: individual build steps show CACHED for RUN/COPY layers, but the final image still
gets a new ID. The exporting steps (exporting layers, writing image, naming) take
0.1–0.2s instead of 0.0s, suggesting they actually re-export rather than reuse.
Compose Version
Docker Compose version v5.1.0
Docker Environment
Client: Docker Engine - Community
Version: 29.3.0
Context: default
Debug Mode: false
Plugins:
buildx: Docker Buildx (Docker Inc.)
Version: v0.31.1
Path: /usr/libexec/docker/cli-plugins/docker-buildx
compose: Docker Compose (Docker Inc.)
Version: v5.1.0
Path: /usr/libexec/docker/cli-plugins/docker-compose
Server:
Containers: 8
Running: 7
Paused: 0
Stopped: 1
Images: 8
Server Version: 29.3.0
Storage Driver: overlayfs
driver-type: io.containerd.snapshotter.v1
Logging Driver: local
Cgroup Driver: systemd
Cgroup Version: 2
Plugins:
Volume: local
Network: bridge host ipvlan macvlan null overlay
Log: awslogs fluentd gcplogs gelf journald json-file local splunk syslog
CDI spec directories:
/etc/cdi
/var/run/cdi
Swarm: inactive
Runtimes: io.containerd.runc.v2 runc
Default Runtime: runc
Init Binary: docker-init
containerd version: 301b2dac98f15c27117da5c8af12118a041a31d9
runc version: v1.3.4-0-gd6d73eb8
init version: de40ad0
Security Options:
apparmor
seccomp
Profile: builtin
cgroupns
Kernel Version: 6.8.0-106-generic
Operating System: Ubuntu 24.04.4 LTS
OSType: linux
Architecture: x86_64
CPUs: 4
Total Memory: 7.57GiB
Name: secondary
ID: 091526cd-489b-4f47-acad-33efcea821cb
Docker Root Dir: /var/lib/docker
Debug Mode: false
Experimental: false
Insecure Registries:
::1/128
127.0.0.0/8
Live Restore Enabled: false
Firewall Backend: iptables
Anything else?
Related issues
docker compose builddoes not completely reuse cached layers on immediate rebuild, butdocker builddoes for-mac#7341 — same symptom on macOS Docker Desktop- Layer not cached under
containerdimage store for-mac#7359 — layer not cached under containerd image store (fixed by increasing
defaultKeepStorage, but that does not help here — layers are small)
Workaround
Disable the containerd image store:
{
"features": {
"containerd-snapshotter": false
}
}After changing daemon.json and restarting Docker, existing images are no longer available
(format is incompatible) and need to be rebuilt/re-pulled. Volumes are preserved.