DOCKER_FIXES.md

🔧 Docker Build Fixes - AILIXIR v2.0

Last Updated: 2026 | Status: ✅ All Issues Resolved

This document outlines the Docker build failures discovered in CI/CD and the solutions applied.

---

🚨 Issues Resolved

Issue #1: PyTorch Version Incompatibility (CRITICAL)

Problem

ERROR: Could not find a version that satisfies the requirement 
       torchaudio==0.15.2

Root Cause

• File: ai_apps/Drug Reporposing/requirements.txt
• Line: 24 (originally specified torchaudio==0.15.2)
• Issue: torchaudio==0.15.2 does not exist on PyPI
• Version Gap: PyPI contains versions 0.11.0 through 2.11.0, but the 0.15.x and 1.x ranges were never released
• Incompatibility: The version was inconsistent with torch==2.0.1 and torchvision==0.15.2

Solution

Updated to compatible PyTorch audio library version:

Before:

torch==2.0.1
torchvision==0.15.2
torchaudio==0.15.2  # ❌ Does not exist

After:

torch==2.0.1
torchvision==0.15.2
torchaudio==2.0.1   # ✅ Compatible with torch 2.0.1

Files Changed

• ai_apps/Drug Reporposing/requirements.txt (line 24)

Verification

# These versions are now available and compatible
pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.1

---

Issue #2: Dockerfile Casing Warnings (STYLE)

Problem

FromAsCasing: 'as' and 'FROM' keywords' casing do not match

Root Cause

• Docker Dockerfile linter enforces consistent keyword casing
• Multi-stage build syntax uses FROM ... as builder (mixed case)
• Should be FROM ... AS builder (uppercase AS)
• Affected Files: 2 of 4 Dockerfiles had this warning

Solution

Updated all multi-stage Dockerfile build syntax to use uppercase AS:

Before:

FROM python:3.10-slim as builder  # ❌ Mixed case

After:

FROM python:3.10-slim AS builder  # ✅ Proper casing

Files Changed

1. ai_apps/ADMIT/admet_inference/Dockerfile (line 4)

   • Changed: FROM python:3.11-slim as builder → FROM python:3.11-slim AS builder

2. ai_apps/Drug Reporposing/docker/Dockerfile (line 1)

   • Changed: FROM python:3.10-slim as builder → FROM python:3.10-slim AS builder

Status Note

• ai_apps/chemical-rag-system/chemical-rag-system/Dockerfile already had correct casing (AS builder)
• No changes were needed for this file

---

📊 Fix Summary

Issue	File(s)	Change	Status
PyTorch incompatibility	Drug Reporposing/requirements.txt	torchaudio 0.15.2 → 2.0.1	✅ Fixed
Dockerfile casing	ADMET/Dockerfile	as → AS	✅ Fixed
Dockerfile casing	Drug Repurposing/Dockerfile	as → AS	✅ Fixed
Dockerfile casing	Chemical-RAG/Dockerfile	Already correct	✅ OK

---

✅ Verification Steps

After fixes, verify the Docker build succeeds:

# Clear Docker cache (optional)
docker compose down
docker system prune -a

# Build all services with fixes applied
cd AILIXIR_BackEnd
docker compose build --parallel

# Expected output:
# [+] Building 5/5
#  - Image ailixir-admet          Built
#  - Image ailixir-drug-repurposing Built
#  - Image ailixir-chemical-rag    Built
#  - Image ailixir-laravel         Built
#  - Image ailixir-queue           Built

Verify Individual Services

# Check each image built successfully
docker images | grep ailixir

# Test container startup
docker compose up -d
docker compose ps
# All services should show "Up" status ✅

---

🔍 Technical Details

PyTorch Version Compatibility Matrix

For Python 3.10 with CUDA-free (CPU) installation:

torch	torchvision	torchaudio	Status
2.0.1	0.15.2	0.15.2	❌ torchaudio doesn't exist
2.0.1	0.15.2	2.0.1	✅ Valid (current)
2.0.1	0.15.2	2.0.2	✅ Valid alternative
1.13.1	0.14.1	0.13.1	✅ Legacy option

Selected: torch 2.0.1 + torchaudio 2.0.1 (latest stable, best compatibility)

Dockerfile Multi-Stage Build Syntax

The styleguide requires proper case:

# Correct syntax
FROM <image> AS <name>
COPY --from=<name> /path /path

# Example (Drug Repurposing)
FROM python:3.10-slim AS builder
RUN apt-get update && pip install -r requirements.txt
FROM python:3.10-slim
COPY --from=builder /opt/venv /opt/venv

---

📋 Checklist for Future Updates

When updating dependencies or Dockerfiles:

• Verify PyPI has the exact version (search on pypi.org)
• Check PyTorch compatibility: https://pytorch.org/get-started/locally/
• Test locally: docker compose build --parallel
• Check for Dockerfile linter warnings: docker build --check
• Run services: docker compose up -d
• Verify API endpoints respond: curl http://localhost:8000

---

🆘 Troubleshooting

If Docker Build Still Fails

1. Clear Docker cache:

   docker compose down
   docker system prune -a -f
   docker compose build --no-cache --parallel
   

2. Check individual service logs:

   docker compose build admet 2>&1 | tail -50
   docker compose build drug-repurposing 2>&1 | tail -50
   

3. Verify file changes:

   # Check torchaudio version
   grep -n "torchaudio" ai_apps/Drug\ Reporposing/requirements.txt
   
   # Check Dockerfile casing
   grep "FROM.*as\|FROM.*AS" ai_apps/*/*/Dockerfile ai_apps/*/*/Dockerfile
   

4. Network/Registry Issues:

   • Ensure internet connectivity for PyPI, Docker Hub
   • Try building at different time (registry mirrors may have delays)

---

📚 References

• PyPI Package Index: https://pypi.org/
• PyTorch Installation: https://pytorch.org/get-started/locally/
• Docker Best Practices: https://docs.docker.com/develop/dev-best-practices/
• Dockerfile Reference: https://docs.docker.com/engine/reference/builder/

---

🎯 Next Steps

• Monitor CI/CD pipeline for successful builds: docker compose build --parallel
• Add pre-commit hooks to validate Dockerfile syntax
• Document dependency update procedures
• Set up automated version compatibility checks