NHP ATS Backup

Back up and restore NHP Azure Table Storage (ATS) via JSON snapshots in Blob Storage.

Prerequisites

• uv
• Azure CLI
• Azure Functions Core Tools (optional, for local testing)

Configuration

Create a .env file in the repo root by copying and filling in the values below:

# Azure resources
AZURE_STORAGE_ACCOUNT_NAME=<storage-name>      # Globally unique, lowercase letters+numbers, 3-24 chars
AZURE_RESOURCE_GROUP_NAME=<resource-group>     # Existing resource group where resources are created
AZURE_LOCATION=<region>                        # e.g. australiaeast; pick one near your users

# Storage configuration
AZURE_SKU=Standard_LRS                         # Cheapest; fine for dev/test and small prod workloads
AZURE_STORAGE_KIND=StorageV2                   # General-purpose v2; supports blobs, queues, tables

# Backup targets
PROD_TABLE_NAME=<ats-name>                     # Production Azure Table Storage table to back up
BACKUP_CONTAINER_NAME=<backup-container-name>  # Blob container for JSON snapshots
DEV_TABLE_NAME=<ats-dev-name>                  # Dev/test table for safe restore experiments

# Function App
AZURE_FUNCTION_APP_NAME=<function-app-name>    # Globally unique, used in URLs and deployment
AZURE_PYTHON_VERSION=3.13                      # Max version supported by Azure Functions; 3.14 is still in preview
AZURE_FUNCTION_INSTANCE_MEMORY=2048            # Flex Consumption memory in MB; 2048 is a good default

Windows users should omit export or use set instead. Notice that AZURE_SKU and AZURE_STORAGE_KIND default to Standard_LRS and StorageV2 if omitted.

Quick deploy

Automated deploy

Run uv run deploy.py to interactively create the resource group, storage account, Function App, backup container, app settings, and publish the functions. Add --yes to skip confirmation prompts.

This project uses Azure Functions on Flex Consumption (Python 3.13).

1. Generate requirements.txt

Azure's remote build uses pip, not uv. Compile a lockfile before every deploy:

uv pip compile pyproject.toml -o requirements.txt

2. Prepare .funcignore

Ensure .funcignore exists so the publish step skips your local venv and build artefacts:

.venv
__pycache__
.git
.env
local.settings.json
.ruff_cache
.pytest_cache
tests/
*.egg-info
*.pyc
.github/

3. Azure resources and app settings

Click to expand: create storage account, Function App, and configure app settings

Create the storage account. The name must be globally unique, lowercase letters and numbers only, 3–24 characters.

az storage account create \
  --name "$AZURE_STORAGE_ACCOUNT_NAME" \
  --resource-group "$AZURE_RESOURCE_GROUP_NAME" \
  --location "$AZURE_LOCATION" \
  --sku "$AZURE_SKU" \
  --kind "$AZURE_STORAGE_KIND"

Create the backup container:

az storage container create \
  --name "$BACKUP_CONTAINER_NAME" \
  --account-name "$AZURE_STORAGE_ACCOUNT_NAME" \
  --auth-mode login

Create the Function App:

az functionapp create \
  --resource-group "$AZURE_RESOURCE_GROUP_NAME" \
  --name "$AZURE_FUNCTION_APP_NAME" \
  --storage-account "$AZURE_STORAGE_ACCOUNT_NAME" \
  --flexconsumption-location "$AZURE_LOCATION" \
  --runtime python \
  --runtime-version "$AZURE_PYTHON_VERSION" \
  --functions-version 4 \
  --instance-memory "$AZURE_FUNCTION_INSTANCE_MEMORY"

Configure app settings:

az functionapp config appsettings set \
  --name "$AZURE_FUNCTION_APP_NAME" \
  --resource-group "$AZURE_RESOURCE_GROUP_NAME" \
  --settings \
    "AZURE_STORAGE_ACCOUNT_NAME=$AZURE_STORAGE_ACCOUNT_NAME" \
    "PROD_TABLE_NAME=$PROD_TABLE_NAME" \
    "BACKUP_CONTAINER_NAME=$BACKUP_CONTAINER_NAME" \
    "DEV_TABLE_NAME=$DEV_TABLE_NAME"

4. Publish

func azure functionapp publish "$AZURE_FUNCTION_APP_NAME"

The app contains two functions:

Function	Trigger	Route / Schedule
nhp_ats_backup	Timer	0 0 2 * * * (02:00 UTC daily)
nhp_ats_backup_dev	HTTP	POST /api/nhp-ats-backup-dev

Usage

Create a backup (local)

uv run --env-file .env python -m backup.core

Restore (interactive)

uv run --env-file .env python -m backup.cli

Restore a specific date

uv run --env-file .env python -m backup.cli --restore-date 2026-06-17

Restore to a specific table

uv run --env-file .env python -m backup.core --restore snapshot.json --target-table <table_name>

Non-interactive restore

# Identify latest snapshot
SNAPSHOT=$(az storage blob list --account-name "$AZURE_STORAGE_ACCOUNT_NAME" \
  --container-name "$BACKUP_CONTAINER_NAME" --query "[].name" -o tsv \
  | grep -E '^\d{4}-' | sort | tail -1)

# Download
az storage blob download --account-name "$AZURE_STORAGE_ACCOUNT_NAME" \
  --container-name "$BACKUP_CONTAINER_NAME" --name "$SNAPSHOT" \
  --file snapshot.json

# Restore
uv run --env-file .env python -m backup.core --restore snapshot.json --target-table <table_name>

Local testing

Click to expand

Create local.settings.json:

import json
import re

env = {}
with open(".env") as f:
    for line in f:
        line = line.strip()
        if line and not line.startswith("#"):
            key, _, value = line.partition("=")
            env[key.strip()] = value.strip().strip("'\"")

settings = {
    "IsEncrypted": False,
    "Values": {
        "AzureWebJobsStorage": "UseDevelopmentStorage=true",
        "FUNCTIONS_WORKER_RUNTIME": "python",
        "AZURE_FUNCTIONS_ENVIRONMENT": "Development",
        "AZURE_STORAGE_ACCOUNT_NAME": env["AZURE_STORAGE_ACCOUNT_NAME"],
        "PROD_TABLE_NAME": env["PROD_TABLE_NAME"],
        "BACKUP_CONTAINER_NAME": env["BACKUP_CONTAINER_NAME"],
        "DEV_TABLE_NAME": env["DEV_TABLE_NAME"],
    }
}

with open("local.settings.json", "w") as f:
    json.dump(settings, f, indent=2)
    f.write("\n")

Save the above Python code in local_settings.py and run it: uv run local_settings.py

Run the host:

func start

Invoke the timer function manually:

curl http://localhost:7071/admin/functions/nhp_ats_backup -X POST -d '{}'

Developer notes

Click to expand

• backup/core.py — library functions and non-interactive backup/restore
• backup/cli.py — interactive workflow and snapshot resolution
• function_app.py — Azure Function entrypoint (timer + HTTP)

Snapshots

• JSON format with EDM type tags for perfect round-trip fidelity
• Stored as YYYY-MM-DDTHH:MMZ.json
• Retention: 7 daily snapshots + 6 monthly keepers (see _prune_snapshots in core.py)

Tests

uv run pytest

Lint / Format

uv run ruff check .
uv run ruff format .

Status reporting

The backup script emits structured logs (success/failure, entity count, latest snapshot) for consumption by nhp_ats_tui.