Rabbitmq 什么会导致“whereis(user)”返回未定义?
我们正在生产系统上运行RabbitMQ 3.6.5,在Rabbitmq 什么会导致“whereis(user)”返回未定义?,rabbitmq,erlang,Rabbitmq,Erlang,我们正在生产系统上运行RabbitMQ 3.6.5,在 Erlang R16B03-1 (erts-5.10.4) [source] [64-bit] [smp:2:2] [async-threads:10] [hipe] [kernel-poll:false] rabbitmqctl rotate_日志出现以下故障: Erlang VM I/O system is damaged, restart needed 查看该版本的代码,此错误仅存在于rabbit\u log.erl: ...
Erlang R16B03-1 (erts-5.10.4) [source] [64-bit] [smp:2:2] [async-threads:10] [hipe] [kernel-poll:false]
rabbitmqctl rotate_日志
出现以下故障:
Erlang VM I/O system is damaged, restart needed
查看该版本的代码,此错误仅存在于rabbit\u log.erl
:
...
%% Execute Fun using the IO system of the local node (i.e. the node on
%% which the code is executing). Since this is invoked for every log
%% message, we try to avoid unnecessarily churning group_leader/1.
with_local_io(Fun) ->
GL = group_leader(),
Node = node(),
case node(GL) of
Node -> Fun();
_ -> set_group_leader_to_user_safely(whereis(user)),
try
Fun()
after
group_leader(GL, self())
end
end.
set_group_leader_to_user_safely(undefined) ->
handle_damaged_io_system();
set_group_leader_to_user_safely(User) when is_pid(User) ->
group_leader(User, self()).
handle_damaged_io_system() ->
Msg = "Erlang VM I/O system is damaged, restart needed~n",
io:format(standard_error, Msg, []),
exit(erlang_vm_restart_needed).
因此,似乎whereis(user)
正在返回undefined
这个错误在运行相同版本的测试系统中没有发生,这意味着到目前为止我无法重现这个错误
在非生产系统中,我可以尝试纠正措施。由于故障发生在生产系统中,因此关键目标是
以尽可能少的中断方式更正此问题
我希望了解这种情况是否会阻止兔子再次启动,以及
重新启动RabbitMQ将更正此问题
日志启动\u err
包含:
Erlang VM I/O system is damaged, restart needed
=SUPERVISOR REPORT==== 28-Nov-2018::18:45:04 ===
Supervisor: {<0.26938.6080>,rabbit_channel_sup}
Context: shutdown_error
Reason: noproc
Offender: [{pid,<0.26217.6080>},
{name,channel},
{mfargs,
{rabbit_channel,start_link,
[1,<0.27525.6080>,<0.25586.6080>,<0.27525.6080>,
<<"40.113.233.192:3979 -> 10.0.0.4:5672">>,
rabbit_framing_amqp_0_9_1,
{user,<<"FDLMessaging">>,[],
[{rabbit_auth_backend_internal,none}]},
<<"/">>,
[{<<"publisher_confirms">>,bool,true},
{<<"exchange_exchange_bindings">>,bool,true},
{<<"basic.nack">>,bool,true},
{<<"consumer_cancel_notify">>,bool,true},
{<<"connection.blocked">>,bool,true},
{<<"authentication_failure_close">>,bool,true}],
<0.26295.6080>,<0.22812.6080>]}},
{restart_type,intrinsic},
{shutdown,70000},
{child_type,worker}]
日志rabbit@fdlquevm-sasl.log-20180710包含:
Erlang VM I/O system is damaged, restart needed
=SUPERVISOR REPORT==== 28-Nov-2018::18:45:04 ===
Supervisor: {<0.26938.6080>,rabbit_channel_sup}
Context: shutdown_error
Reason: noproc
Offender: [{pid,<0.26217.6080>},
{name,channel},
{mfargs,
{rabbit_channel,start_link,
[1,<0.27525.6080>,<0.25586.6080>,<0.27525.6080>,
<<"40.113.233.192:3979 -> 10.0.0.4:5672">>,
rabbit_framing_amqp_0_9_1,
{user,<<"FDLMessaging">>,[],
[{rabbit_auth_backend_internal,none}]},
<<"/">>,
[{<<"publisher_confirms">>,bool,true},
{<<"exchange_exchange_bindings">>,bool,true},
{<<"basic.nack">>,bool,true},
{<<"consumer_cancel_notify">>,bool,true},
{<<"connection.blocked">>,bool,true},
{<<"authentication_failure_close">>,bool,true}],
<0.26295.6080>,<0.22812.6080>]}},
{restart_type,intrinsic},
{shutdown,70000},
{child_type,worker}]
=主管报告===2018年11月28日::18:45:04===
主管:{,兔子频道}
上下文:关机错误
原因:noproc
罪犯:[{pid,},
{name,channel},
{mfargs,
{兔子频道,启动链接,
[1,,,,
>,
rabbit_framing_amqp_0_9_1,
{用户,,[],
[{rabbit_auth_backend_internal,none}]},
,
[{,bool,true},
{,bool,true},
{,bool,true},
{,bool,true},
{,bool,true},
{,bool,true}],
,]}},
{restart_type,injective},
{关闭,70000},
{child_type,worker}]
什么会导致whereis(user)
返回未定义?用户进程是一个中央进程,处理与标准输出相关的大部分I/O。默认情况下,它将始终存在,除非将-nouser
参数提供给底层Erlang虚拟机
这个进程可能失败的另一种方式是,如果它被另一个恶意进程杀死,或者被VM执行类似杀死一个使用过多内存的进程的操作,或者因为它引用了热代码加载过多次的代码,需要清除它
除非有人或某事愿意拨弄该代码,否则基本上不会发生这种情况。您使用的是什么版本的Erlang?RabbitMQ日志中有什么有趣的内容吗?解释用户
过程是什么以及它的作用。请注意,RabbitMQ3.7.X
版本没有此代码,因为日志记录切换到了lager
库。还有一件事,在RabbitMQ核心工程团队监控时,最好询问这类问题。@LukeBakken-在问题中添加了Erlang版本。也谢谢你给rabbitmq用户的链接。@LukeBakken-现在问题也包括日志信息。