refactor: optimize Docker Compose with YAML anchors and aliases

[chris-stinemetz]

• Reduce configuration repetition by ~85% using YAML anchors (&) and aliases (*)
• Extract common patterns into reusable anchors:
  • x-common-config: shared dependencies, env files, and networking
  • x-huggingface-cache: HF cache environment variables (~6 vars per service)
  • x-auth-config: authentication configuration (~8 vars per service)
  • x-embedding-config: embedding model settings (~4 vars per service)
  • x-reranker-config: reranker settings (~7 vars per service)
  • x-common-volumes & x-indexer-volumes: volume mount patterns
• Eliminate ~200+ lines of repetitive environment variable declarations
• Improve maintainability with single source of truth for shared configs
• Maintain full functionality across all services (validated with deployment test)

Files optimized:

• docker-compose.yml: 8 services now use shared anchors
• docker-compose.openlit.yml: health check dependency pattern
• docker-compose-bindmount-checkout.yml: working_dir and common configs

[augmentcode]

🤖 Augment PR Summary

Summary: Refactors the project’s Docker Compose configurations to reduce repetition by introducing reusable YAML anchors/aliases.

Changes:

• Added shared extension blocks (e.g., common env/depends_on, HuggingFace cache vars, auth vars, embedding vars, reranker vars, and common volume patterns).
• Updated multiple services to consume these anchors via YAML merge keys, removing large duplicated environment/volume sections.
• Refactored healthcheck/depends_on patterns in the OpenLit overlay compose file using an anchor for the ClickHouse health condition.
• Normalized several array-style fields (e.g., command/entrypoint/healthcheck.test) for consistency while preserving intent.
• Adjusted the llamacpp service configuration in docker-compose.yml (context size, GPU layer args, and resource settings).

Technical Notes: Uses Compose extension fields (x-*) with YAML anchors to centralize configuration and simplify future maintenance; verify Compose version compatibility with YAML merge usage and any runtime-specific handling of resource constraints.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 2 suggestions posted.

Comment augment review to trigger a new review at any time.

[augmentcode]

llamacpp now sets LLAMA_ARG_CTX_SIZE=4096 (was 8192) and the command forces --n-gpu-layers 0, which is a functional behavior change beyond a pure YAML refactor. Can you confirm this is intentional and captured in the PR description since it can affect output quality/perf/memory?

_{🤖 Was this useful? React with 👍 or 👎}

[chris-stinemetz]

I changed to default 8192 and allow override via .env file.

[augmentcode]

FYI: deploy.resources is ignored by docker compose up in many non-Swarm setups, so these memory limits/reservations may not actually be enforced. Can you confirm the target runtime will honor this so it doesn’t give a false sense of protection against OOMs?

_{🤖 Was this useful? React with 👍 or 👎}