(26.10) Installation Test: openIMIS with Docker | Windows
Contents
Test Information
Date | 2026-11-04 |
|---|---|
Tester | Zaki Syaiful Hunafa |
openIMIS release | Release 2026-10 Release |
Result | PASS |
Comment |
|
Technical specifications
Operating System | Windows 11 25H2 |
|---|---|
Hardware Type | Laptop |
Hardware Model | MSI GF63 Thin 10SCSR |
Processor | Intel Core i5-10200H |
RAM | 16GB |
Secondary Storage | 500GB |
Docker | YES |
Activity log
Installed with Docker
Results
Confirmed functionality of frontend and backend
Observations/Lessons Learned
The openIMIS Docker deployment (distribution version 25.10) was installed successfully on a Windows machine running Docker Desktop. The installation did not work on the first attempt. A Redis container failed its health check, which blocked the dependent services, and resolving it took several rounds of diagnosis. The error messages shown during startup did not point directly to the cause.
What Happened
1. First startup
docker compose up -d pulled all images successfully. Two warnings appeared (OPENIMIS_FE_CONF_JSON not set, and deploy.mode only honored in Swarm mode). Neither blocked the installation.
The Redis container was reported as unhealthy, and Compose stopped with dependency failed to start: container ...redis-1 is unhealthy. The migrations container, workers and backend were left waiting on Redis. The first failure took 63.1 seconds to report; later attempts failed after about 7 seconds.
2. Locating the cause
The Compose output did not state why Redis was unhealthy. The cause only appeared in the container logs:
exec /usr/local/bin/docker-entrypoint.sh: exec format error
The container was in a restart loop with exit code 255.
3. Checks performed
Image architecture and Docker engine architecture were both
linux/amd64, andDOCKER_DEFAULT_PLATFORMwas not set. An architecture mismatch, the usual meaning of this error, was therefore ruled out.redis:7was pulled and ran correctly when started directly withdocker run(Redis server v7.4.12), and its shell worked.The Compose file was still using
redis:latest, so the pulledredis:7was not the image under test.docker compose configshowed the Redis service as defined incompose.cache.yml: the standard entrypoint, aredis-servercommand with a password, a named data volume, a bind-mountedredis.conf, and aredis-clihealth check. No custom entrypoint or script override was found.python3is not available in the Git Bash environment, so one suggested diagnostic command could not run.
4. Docker engine errors
During a later pull of the large images (OpenSearch is over 1 GB), the pull failed with failed to copy: ... input/output error while writing to Docker Desktop's content store. Subsequent commands then returned 500 Internal Server Error on the Docker API, followed by failed to connect to the docker API ... The system cannot find the file specified, indicating the Docker engine had stopped.
5. Resolution
After Docker Desktop was restarted, the Redis image in compose.cache.yml was changed from redis:latest to redis:7, the stale redis:latest image was removed, and images were re-pulled. Redis then started and became healthy, and the stack came up. The user confirmed it was working.
Likely Cause
The redis:latest image in the local Docker image store was most likely damaged, which is consistent with the input/output errors seen during image pulls. This was not independently confirmed, for example by inspecting the image layers.
Observations on Ease of Installation
The installation succeeded, but it was not straightforward.
The failure surfaced as a generic "unhealthy" dependency error, with no indication of the underlying problem. The actual error was visible only through
docker logs.The
exec format errormessage normally suggests a CPU architecture mismatch, which was not the case here, so it pointed toward a wrong cause.The Redis service is defined in a separate included file (
compose.cache.yml), so finding its configuration required reading through several included Compose files.The Redis image is referenced as
latest(unpinned), and the working configuration required a specific version (redis:7).Two startup warnings appear on every run even though they do not affect operation.
Reaching a working state required roughly ten diagnostic steps and a Docker Desktop restart.