本文へ移動
cccskills
無料GitHub で公開

azure-servicebus-py

Azure Service Bus SDK for Python messaging. Use for queues, topics, subscriptions, and enterprise messaging patterns. Triggers: "service bus", "ServiceBusClient", "queue", "topic", "subscription", "message broker".

インストール方法を見る

含まれるファイル(4)

  • SKILL.md10.1 KB
  • references/dead-letter.md13.8 KB
  • references/patterns.md13.6 KB
  • scripts/setup_servicebus.py13.0 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Azure Service Bus SDK for Python

Enterprise messaging for reliable cloud communication with queues and pub/sub topics.

Installation

pip install azure-servicebus azure-identity

Environment Variables

SERVICEBUS_FULLY_QUALIFIED_NAMESPACE=<namespace>.servicebus.windows.net  # Required for all auth methods
SERVICEBUS_QUEUE_NAME=myqueue  # Required for queue operations
SERVICEBUS_TOPIC_NAME=mytopic  # Required for topic operations
SERVICEBUS_SUBSCRIPTION_NAME=mysubscription  # Required for subscription operations
AZURE_TOKEN_CREDENTIALS=prod # Required only if DefaultAzureCredential is used in production

Authentication & Lifecycle

🔑 Two rules apply to every code sample below:

  1. Prefer DefaultAzureCredential. It works locally (Azure CLI / VS Code / Developer CLI) and in Azure (managed identity, workload identity) with no code change. Avoid connection strings, account/API keys — they bypass Entra audit and rotation.
    • Local dev: DefaultAzureCredential works as-is.
    • Production: set AZURE_TOKEN_CREDENTIALS=prod (or AZURE_TOKEN_CREDENTIALS=<specific_credential>) to constrain the credential chain to production-safe credentials.
  2. Wrap every client in a context manager so HTTP transports, sockets, and token caches are released deterministically:
    • Sync: with <Client>(...) as client:
    • Async: async with <Client>(...) as client: and async with DefaultAzureCredential() as credential: (from azure.identity.aio)

Snippets may abbreviate this setup, but production code should always follow both rules.

from azure.identity import DefaultAzureCredential, ManagedIdentityCredential
from azure.servicebus import ServiceBusClient

# Local dev: DefaultAzureCredential. Production: set AZURE_TOKEN_CREDENTIALS=prod or AZURE_TOKEN_CREDENTIALS=<specific_credential>
credential = DefaultAzureCredential(require_envvar=True)
# Or use a specific credential directly in production:
# See https://learn.microsoft.com/python/api/overview/azure/identity-readme?view=azure-python#credential-classes
# credential = ManagedIdentityCredential()
namespace = "<namespace>.servicebus.windows.net"

with ServiceBusClient(
    fully_qualified_namespace=namespace,
    credential=credential
) as client:
    # Use client here (see following sections for operations)
    ...

Client Types

ClientPurposeGet From
ServiceBusClientConnection managementDirect instantiation
ServiceBusSenderSend messagesclient.get_queue_sender() / get_topic_sender()
ServiceBusReceiverReceive messagesclient.get_queue_receiver() / get_subscription_receiver()

Send Messages (Async)

import asyncio
from azure.servicebus.aio import ServiceBusClient
from azure.servicebus import ServiceBusMessage
from azure.identity.aio import DefaultAzureCredential

async def send_messages():
    credential = DefaultAzureCredential()
    
    async with ServiceBusClient(
        fully_qualified_namespace="<namespace>.servicebus.windows.net",
        credential=credential
    ) as client:
        sender = client.get_queue_sender(queue_name="myqueue")
        
        async with sender:
            # Single message
            message = ServiceBusMessage("Hello, Service Bus!")
            await sender.send_messages(message)
            
            # Batch of messages
            messages = [ServiceBusMessage(f"Message {i}") for i in range(10)]
            await sender.send_messages(messages)
            
            # Message batch (for size control)
            batch = await sender.create_message_batch()
            for i in range(100):
                try:
                    batch.add_message(ServiceBusMessage(f"Batch message {i}"))
                except ValueError:  # Batch full
                    await sender.send_messages(batch)
                    batch = await sender.create_message_batch()
                    batch.add_message(ServiceBusMessage(f"Batch message {i}"))
            await sender.send_messages(batch)

asyncio.run(send_messages())

Receive Messages (Async)

async def receive_messages():
    credential = DefaultAzureCredential()
    
    async with ServiceBusClient(
        fully_qualified_namespace="<namespace>.servicebus.windows.net",
        credential=credential
    ) as client:
        receiver = client.get_queue_receiver(queue_name="myqueue")
        
        async with receiver:
            # Receive batch
            messages = await receiver.receive_messages(
                max_message_count=10,
                max_wait_time=5  # seconds
            )
            
            for msg in messages:
                print(f"Received: {str(msg)}")
                await receiver.complete_message(msg)  # Remove from queue

asyncio.run(receive_messages())

Receive Modes

ModeBehaviorUse Case
PEEK_LOCK (default)Message locked, must complete/abandonReliable processing
RECEIVE_AND_DELETERemoved immediately on receiveAt-most-once delivery
from azure.servicebus import ServiceBusReceiveMode

receiver = client.get_queue_receiver(
    queue_name="myqueue",
    receive_mode=ServiceBusReceiveMode.RECEIVE_AND_DELETE
)

Message Settlement

async with receiver:
    messages = await receiver.receive_messages(max_message_count=1)
    
    for msg in messages:
        try:
            # Process message...
            await receiver.complete_message(msg)  # Success - remove from queue
        except ProcessingError:
            await receiver.abandon_message(msg)  # Retry later
        except PermanentError:
            await receiver.dead_letter_message(
                msg,
                reason="ProcessingFailed",
                error_description="Could not process"
            )
ActionEffect
complete_message()Remove from queue (success)
abandon_message()Release lock, retry immediately
dead_letter_message()Move to dead-letter queue
defer_message()Set aside, receive by sequence number

Topics and Subscriptions

# Send to topic
sender = client.get_topic_sender(topic_name="mytopic")
async with sender:
    await sender.send_messages(ServiceBusMessage("Topic message"))

# Receive from subscription
receiver = client.get_subscription_receiver(
    topic_name="mytopic",
    subscription_name="mysubscription"
)
async with receiver:
    messages = await receiver.receive_messages(max_message_count=10)

Sessions (FIFO)

# Send with session
message = ServiceBusMessage("Session message")
message.session_id = "order-123"
await sender.send_messages(message)

# Receive from specific session
receiver = client.get_queue_receiver(
    queue_name="session-queue",
    session_id="order-123"
)

# Receive from next available session
from azure.servicebus import NEXT_AVAILABLE_SESSION
receiver = client.get_queue_receiver(
    queue_name="session-queue",
    session_id=NEXT_AVAILABLE_SESSION
)

Scheduled Messages

from datetime import datetime, timedelta, timezone

message = ServiceBusMessage("Scheduled message")
scheduled_time = datetime.now(timezone.utc) + timedelta(minutes=10)

# Schedule message
sequence_number = await sender.schedule_messages(message, scheduled_time)

# Cancel scheduled message
await sender.cancel_scheduled_messages(sequence_number)

Dead-Letter Queue

from azure.servicebus import ServiceBusSubQueue

# Receive from dead-letter queue
dlq_receiver = client.get_queue_receiver(
    queue_name="myqueue",
    sub_queue=ServiceBusSubQueue.DEAD_LETTER
)

async with dlq_receiver:
    messages = await dlq_receiver.receive_messages(max_message_count=10)
    for msg in messages:
        print(f"Dead-lettered: {msg.dead_letter_reason}")
        await dlq_receiver.complete_message(msg)

Sync Client (for simple scripts)

from azure.servicebus import ServiceBusClient, ServiceBusMessage
from azure.identity import DefaultAzureCredential

with ServiceBusClient(
    fully_qualified_namespace="<namespace>.servicebus.windows.net",
    credential=DefaultAzureCredential()
) as client:
    with client.get_queue_sender("myqueue") as sender:
        sender.send_messages(ServiceBusMessage("Sync message"))
    
    with client.get_queue_receiver("myqueue") as receiver:
        for msg in receiver:
            print(str(msg))
            receiver.complete_message(msg)

Best Practices

  1. Pick sync OR async and stay consistent. Do not mix azure.xxx sync clients with azure.xxx.aio async clients in the same call path. Choose one mode per module.
  2. Always use context managers for clients and async credentials. Wrap every client in with Client(...) as client: (sync) or async with Client(...) as client: (async) for proper cleanup. For async DefaultAzureCredential from azure.identity.aio, also use async with credential: so tokens and transports are cleaned up.
  3. Use DefaultAzureCredential for portable auth across local dev and Azure (avoid connection strings / API keys when possible).
  4. Use async client for production workloads
  5. Complete messages after successful processing
  6. Use dead-letter queue for poison messages
  7. Use sessions for ordered, FIFO processing
  8. Use message batches for high-throughput scenarios
  9. Set max_wait_time to avoid infinite blocking

Reference Files

FileContents
references/patterns.mdCompeting consumers, sessions, retry patterns, request-response, transactions
references/dead-letter.mdDLQ handling, poison messages, reprocessing strategies
scripts/setup_servicebus.pyCLI for queue/topic/subscription management and DLQ monitoring

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Build Azure AI Foundry agents using the Microsoft Agent Framework Python SDK (agent-framework-azure-ai). Use when creating persistent agents with AzureAIAgentsProvider, using hosted tools (code interpreter, file search, web search), integrating MCP servers, managing conversation threads, or implementing streaming responses. Covers function tools, structured outputs, and multi-tool agents.

日本語の概要は準備中です。原文の説明を表示しています。

microsoft/skills3,1052026年10月13日 更新

Set up AI Runway on AKS — from bare cluster to running model. Covers cluster verification, controller install, GPU assessment, provider setup, and first deployment. WHEN: "setup AI Runway", "onboard AKS cluster", "install AI Runway", "airunway setup", "deploy model to AKS", "GPU inference on AKS", "KAITO setup on AKS", "run LLM on AKS", "vLLM on AKS", "set up model serving on AKS", "AI Runway controller".

日本語の概要は準備中です。原文の説明を表示しています。

microsoft/skills3,1052026年10月13日 更新

Diagnose Day-2 AKS GPU and KAITO incidents using profile-aware, read-only evidence. WHEN: 'Insufficient nvidia.com/gpu', GPU pod Pending, model-load OOM, DCGM/VRAM, KAITO Workspace not ready, or GPU autoscaling. DO NOT USE FOR: setup (airunway-aks-setup), non-GPU incidents (aks-troubleshooting), standalone VM quota (azure-quotas), or generic cost (cost-analysis or cost-optimization from the optional azure-cost plugin).

日本語の概要は準備中です。原文の説明を表示しています。

microsoft/skills3,1052026年10月13日 更新

Lookup documented AKS fixes only when the prompt includes an exact catalog signature and all of its qualifiers: VMCannotFitEphemeralOSDisk; NodePoolMcVersionIncompatible; 'NodeImageVersion is not accepted'; AKS SkuNotAvailable with size, location, and zone; ZonalAllocationFailed with insufficient zone capacity; OverconstrainedAllocationRequest with listed constraints; nested AKS vmssCSE/CSE VMExtensionError_OutboundConnFail, VMExtensionError_K8SAPIServerConnFail, or VMExtensionError_K8SAPIServerDNSLookupFail; or AllocationFailed with the full cataloged internal-error or insufficient-regional-capacity message. Never use for quota errors, code-only or bare wrappers, generic symptoms, incomplete signatures, or failures outside AKS; use aks-troubleshooting or azure-diagnostics.

日本語の概要は準備中です。原文の説明を表示しています。

microsoft/skills3,1052026年10月13日 更新

Collects bounded packet captures from AKS nodes and Azure network configuration for wire-level evidence. WHEN: "capture packets on an AKS node", "take a pcap", "run tcpdump on AKS", "prove where packets drop". Use for explicit packet-capture intent after read-only diagnostics, not general AKS connectivity or ingress troubleshooting.

日本語の概要は準備中です。原文の説明を表示しています。

microsoft/skills3,1052026年10月13日 更新

Debug live Azure Kubernetes Service (AKS) incidents with a read-only, evidence-first investigation. WHEN: pod crashes or Pending, CrashLoopBackOff, OOMKilled, ImagePullBackOff, node NotReady, DNS or ingress failure, connectivity timeout, network policy, SNAT exhaustion, node-pool scaling blocked by QuotaExceeded or InsufficientVCPUQuota, upgrade stuck, spot or zone disruption, a bare VMExtensionProvisioningError or AllocationFailed wrapper, an uncataloged capacity symptom, or 'investigate my AKS cluster'. DO NOT USE FOR: packet capture (use aks-network-capture); GPU or model-serving issues (use aks-gpu-inference); cluster creation or provisioning (use azure-kubernetes); cost (use cost-analysis from the optional azure-cost plugin); pod rightsizing (use azure-kubernetes); a fully qualified documented AKS signature with every required nested qualifier (use aks-known-issues); standalone failures on non-AKS Azure resources (use azure-diagnostics). Unqualified errors and open-ended incidents stay here only for AKS.

日本語の概要は準備中です。原文の説明を表示しています。

microsoft/skills3,1052026年10月13日 更新

microsoft のスキルをすべて見る

このスキルの問題を報告する