[ServiceBus] Send message didn't recovered after network disconnect.

• Package Name: azure-servicebus
• Package Version: 7.0.0b8
• Operating System: Mac 10.15.5
• Python Version: 3.6.10

Describe the bug sending message was infinitely blocked when network interrupted

To Reproduce Steps to reproduce the behavior: with the function from example like this

def send_batch_message(sender):
    batch_message = sender.create_message_batch()
    print('disconnect network here')
    time.sleep(15)
    for _ in range(10):
        try:
            batch_message.add_message(ServiceBusMessage("Session Message inside a ServiceBusMessageBatch", session_id=SESSION_ID))
        except ValueError:
            # ServiceBusMessageBatch object reaches max_size.
            # New ServiceBusMessageBatch object can be created here to send more data.
            break
    sender.send_messages(batch_message)
    print('success')

1. disconnect the network in time.sleep()
2. then wait till it to sender.send_messages
3. reconnect network
4. infinite block

Expected behavior should have mechanism to retry or raise exception at least

[cropse]

I’m using Mac as test operating system, so probably is the specific issue only on macOS.

here is entire code:

#!/usr/bin/env python

import time
from azure.servicebus import ServiceBusClient, ServiceBusMessage

CONNECTION_STR = ""
TOPIC_NAME = ''
SUBSCRIPTION_NAME = ''
SESSION_ID = ''


def send_batch_message(sender):
    batch_message = sender.create_message_batch()
    print('disconnect network here')
    time.sleep(15)
    for _ in range(10):
        try:
            batch_message.add_message(ServiceBusMessage("Session Message inside a ServiceBusMessageBatch", session_id=SESSION_ID))
        except ValueError:
            # ServiceBusMessageBatch object reaches max_size.
            # New ServiceBusMessageBatch object can be created here to send more data.
            break
    print('start to send message, reconnect your network!!')
    sender.send_messages(batch_message)
    print('success !!')


servicebus_client = ServiceBusClient.from_connection_string(conn_str=CONNECTION_STR, logging_enable=False, retry_total=10)
with servicebus_client:
    receiver = servicebus_client.get_subscription_receiver(
        topic_name=TOPIC_NAME,
        subscription_name=SUBSCRIPTION_NAME
    )
    sender = servicebus_client.get_topic_sender(TOPIC_NAME)
    with sender:
        send_batch_message(sender)

print("Receive is done.")

[yunhaoling]

hey @cropse, we have enhanced the underlying C library to be able to detect error IO status on MacOS.

uamqp v1.2.13 is released today to include the fix. Please upgrade to the latest uamqp version (e.g. pip install uamqp --upgrade) to see if it solves your problem.

I’m closing the issue now, feel free to reopen if you still get trouble with the latest version.