1、rabbitmq使用guest用户连接报错，rabbitmq日志报错信息为 "guest" user can only connect via localhost"
rabbitmq官网上有解决方法：
注：在centos系统是可以的，ubuntu下使用guest用户连接是报错的。
2、haproxy代理rabbitmq(openstack HA环境下)，经常报这个错：
。。。 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit (40, 11), # Channel.exchange_declare_ok 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit File "/usr/lib/python2.7/site-packages/amqp/abstract_channel.py", line 67, in wait 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit self.channel_id, allowed_methods) 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit File "/usr/lib/python2.7/site-packages/amqp/connection.py", line 237, in _wait_method 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit self.method_reader.read_method() 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit File "/usr/lib/python2.7/site-packages/amqp/method_framing.py", line 189, in read_method 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit raise m 2015-04-06 20:12:45.187 18618 TRACE oslo.messaging._drivers.impl_rabbit IOError: Socket closed
解决方法：
参考这里 # 查看镜像队列的master队列和slave队列 rabbitmqctl list_queues name pid slave_pids synchronised_slave_pids
3、rabbitmq cluster日志中报handshake timeout
解决方法：增加handshake timeout值
注：rabbitmq 3.3.5代码中handshake_timeout是hard code，写死值是10000毫秒；
详情见 https://github.com/rabbitmq/rabbitmq-server/tree/rabbitmq_v3_3_5 （搜索关键字handshake_timeout）
-export([system_continue/3, system_terminate/4, system_code_change/4]). -export([init/2, mainloop/4, recvloop/4]). -export([conserve_resources/3, server_properties/1]). -define(HANDSHAKE_TIMEOUT, 10). # hard code -define(NORMAL_TIMEOUT, 3). -define(CLOSING_TIMEOUT, 30). -define(CHANNEL_TERMINATION_TIMEOUT, 3). -define(SILENT_CLOSE_DELAY, 3). -define(CHANNEL_MIN, 1).
4、 rabbitmq连接缺少心跳检测或tcp长连接保持（OpenStakc Juno + Rabbitmq 3.3.5），通常是控制节点和计算节点之间存在防火墙会把这个问题放大。
解决方法：
1、修改/etc/rabbitmq/rabbitmq.conf
[
{rabbit, [{tcp_listen_options, [binary,
{packet, raw},
{reuseaddr, true},
{backlog, 128},
{nodelay, true},
{exit_on_close, false},
{keepalive, true}]} # 主要增加这行
]}
].
2、操作系统层面tuning
echo "5" > /proc/sys/net/ipv4/tcp_keepalive_time
echo "5" > /proc/sys/net/ipv4/tcp_keepalive_probes
echo "1" > /proc/sys/net/ipv4/tcp_keepalive_intvl
bug描述: https://bugs.launchpad.net/nova/+bug/856764
OpenStack Juno 增加rabbitmq heartbeat patch: https://review.openstack.org/#/c/126330/
5、rabbitmq集群脑裂问题
针对network partition的分析
http://my.oschina.net/moooofly/blog/424660
6、msg_store_persistent、msg_store_transient概念
[root@controller1 rabbit(keystone_admin)]# du -h /var/lib/rabbitmq/mnesia/rabbit/msg_store_transient/ 48G/var/lib/rabbitmq/mnesia/rabbit/msg_store_transient/
重启rabbitmq-server服务，msg_store_transient目录下的*.rdq文件都删除了。
参考链接：http://fanli7.net/a/qitazonghe/IBM_MQ___TLQ/20140217/470409.html
参考链接
http://www.rabbitmq.com/access-control.html
