Address shutdown behavior

[NNTin]

Analysis of the problem

Problem Analysis

Based on the error traceback and codebase investigation, here's what's happening:

The Error

AttributeError: 'Red' object has no attribute '_AutoShardedClient__queue'. Did you mean: '_AutoShardedClient__shards'?

This error occurs during the bot shutdown process when red.close() is called. The issue is that the Red class (which inherits from discord.ext.commands.AutoShardedBot) is trying to access an internal attribute __queue that doesn't exist.

Root Cause

The problem stems from the mixin initialization order in the Red class:

class Red(
    commands.GroupMixin, RPCMixin, dpy_commands.bot.AutoShardedBot
):

Looking at the initialization chain:

1. Red.init (line 104-241 in bot.py) does extensive setup work BEFORE calling super().init() at line 241

2. RPCMixin.init (line 144-146 in _rpc.py) calls super().init(**kwargs)

3. commands.GroupMixin doesn't define init, so it passes through

4. Finally reaches AutoShardedBot.init which sets up internal attributes like __queue

The issue is that during the --dry-run execution:

• The bot starts initialization

• At line 366 in main.py, when --dry-run is detected, it calls sys.exit(ExitCodes.SHUTDOWN)

• This triggers the shutdown handler at line 539

• The shutdown handler calls await red.close() at line 441

• But red.close() → super().close() (line 2278 in bot.py) tries to access AutoShardedClient's internal __queue attribute

• This attribute was never properly initialized because the initialization was interrupted

Why This Happens Specifically with --dry-run

The --dry-run flag causes an early exit at line 365-366 in main.py:

if cli_flags.dry_run:
    sys.exit(ExitCodes.SHUTDOWN)

This happens before await red.start(token) is called (line 369), which means:

• The bot's connection lifecycle never begins

• Discord.py's internal sharding infrastructure (including __queue) is never fully initialized

• But the shutdown handler still tries to call close() which expects these internals to exist

try, catch might not be the best solution but I can't think of a better one. This only happens when it is not fully initialized and thus sharding is not possible.

closes #6572
the original issue is also addressed.

PS C:\privat\codex\Red-DiscordBot> python -m redbot --no-prompt
Error: No instance name was provided!
PS C:\privat\codex\Red-DiscordBot> python -m redbot --no-prompt tinkerer
Instance with name 'tinkerer' doesn't exist. You can create new instance using `redbot-setup` prior to running the bot.