pyr0ball
26fc97dfe5
Three inter-related fixes for the service adoption flow: - preflight: stub_port field — adopted services get a free port for their no-op container (avoids binding conflict with external service on real port) while update_llm_yaml still uses the real external port for host.docker.internal URLs - preflight: write_env now uses stub_port (not resolved) for adopted services so SEARXNG_PORT etc point to the stub's harmless port, not the occupied one - preflight: stub containers use sleep infinity + CMD true healthcheck so depends_on: service_healthy is satisfied without holding any real port - Makefile: finetune profile changed from [cpu,single-gpu,dual-gpu] to [finetune] so the pytorch/cuda base image is not built during make start
73 lines
2.9 KiB
Makefile
73 lines
2.9 KiB
Makefile
# Makefile — Peregrine convenience targets
|
|
# Usage: make <target>
|
|
|
|
.PHONY: setup preflight start stop restart logs test prepare-training finetune clean help
|
|
|
|
PROFILE ?= remote
|
|
PYTHON ?= python3
|
|
|
|
# Auto-detect container engine: prefer docker compose, fall back to podman
|
|
COMPOSE ?= $(shell \
|
|
command -v docker >/dev/null 2>&1 && docker compose version >/dev/null 2>&1 \
|
|
&& echo "docker compose" \
|
|
|| (command -v podman >/dev/null 2>&1 \
|
|
&& podman compose version >/dev/null 2>&1 \
|
|
&& echo "podman compose" \
|
|
|| echo "podman-compose"))
|
|
|
|
# GPU profiles require an overlay for NVIDIA device reservations.
|
|
# Docker uses deploy.resources (compose.gpu.yml); Podman uses CDI device specs (compose.podman-gpu.yml).
|
|
# Generate CDI spec for Podman first: sudo nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml
|
|
#
|
|
# NOTE: When explicit -f flags are used, Docker Compose does NOT auto-detect
|
|
# compose.override.yml. We must include it explicitly when present.
|
|
OVERRIDE_FILE := $(wildcard compose.override.yml)
|
|
COMPOSE_OVERRIDE := $(if $(OVERRIDE_FILE),-f compose.override.yml,)
|
|
|
|
COMPOSE_FILES := -f compose.yml $(COMPOSE_OVERRIDE)
|
|
ifneq (,$(findstring podman,$(COMPOSE)))
|
|
ifneq (,$(findstring gpu,$(PROFILE)))
|
|
COMPOSE_FILES := -f compose.yml $(COMPOSE_OVERRIDE) -f compose.podman-gpu.yml
|
|
endif
|
|
else
|
|
ifneq (,$(findstring gpu,$(PROFILE)))
|
|
COMPOSE_FILES := -f compose.yml $(COMPOSE_OVERRIDE) -f compose.gpu.yml
|
|
endif
|
|
endif
|
|
|
|
setup: ## Install dependencies (Docker or Podman + NVIDIA toolkit)
|
|
@bash setup.sh
|
|
|
|
preflight: ## Check ports + system resources; write .env
|
|
@$(PYTHON) scripts/preflight.py
|
|
|
|
start: preflight ## Preflight check then start Peregrine (PROFILE=remote|cpu|single-gpu|dual-gpu)
|
|
$(COMPOSE) $(COMPOSE_FILES) --profile $(PROFILE) up -d
|
|
|
|
stop: ## Stop all Peregrine services
|
|
$(COMPOSE) down
|
|
|
|
restart: preflight ## Preflight check then restart all services
|
|
$(COMPOSE) down && $(COMPOSE) $(COMPOSE_FILES) --profile $(PROFILE) up -d
|
|
|
|
logs: ## Tail app logs
|
|
$(COMPOSE) logs -f app
|
|
|
|
test: ## Run the test suite
|
|
@$(PYTHON) -m pytest tests/ -v
|
|
|
|
prepare-training: ## Scan docs_dir for cover letters and build training JSONL
|
|
$(COMPOSE) $(COMPOSE_FILES) run --rm app python scripts/prepare_training_data.py
|
|
|
|
finetune: ## Fine-tune your personal cover letter model (run prepare-training first)
|
|
@echo "Starting fine-tune (30-90 min on GPU, much longer on CPU)..."
|
|
$(COMPOSE) $(COMPOSE_FILES) -f compose.gpu.yml --profile finetune run --rm finetune
|
|
|
|
clean: ## Remove containers, images, and data volumes (DESTRUCTIVE)
|
|
@echo "WARNING: This will delete all Peregrine containers and data."
|
|
@read -p "Type 'yes' to confirm: " confirm && [ "$$confirm" = "yes" ]
|
|
$(COMPOSE) down --rmi local --volumes
|
|
|
|
help: ## Show this help
|
|
@grep -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | \
|
|
awk 'BEGIN {FS = ":.*?## "}; {printf " \033[36m%-12s\033[0m %s\n", $$1, $$2}'
|