The process of running RabbitMQ on Linux

Introduction

Running a RabbitMQ server with a deeper look looks like a very complicated affair. Why this is and how everything is actually can be read under the cut.

1. Version RabbitMQ-Server - 2.1.0
2. OS: Fedora

Short

Carefully look at the PIDs of the processes. Below is an explanation of what is happening:

 -bash-3.2# /etc/init.d/rabbitmq-server start root 7732 3834 0 11:54 pts/0 00:00:00 /bin/sh /etc/init.d/rabbitmq-server start root 7733 7732 0 11:54 pts/0 00:00:00 /bin/sh /usr/sbin/rabbitmq-multi start_all 1 root 7740 7733 0 11:54 pts/0 00:00:00 su rabbitmq -s /bin/sh -c /usr/lib/rabbitmq/bin/rabbitmq-multi "start_all" "1" rabbitmq 7741 7740 7 11:54 ? 00:00:00 /usr/lib64/erlang/erts-5.7.5/bin/beam.smp -- -root /usr/lib64/erlang -progname erl -- -home /var/lib/rabbitmq -- -pa /usr/lib/rabbitmq/lib/rabbitmq_server- 2.1.0/sbin/../ebin -noshell -noinput -hidden -K true -sname rabbitmq_multi7741 -s rabbit_multi -extra start_all 1 rabbitmq 7767 7741 0 11:54 ? 00:00:00 inet_gethost 4 rabbitmq 7768 7767 0 11:54 ? 00:00:00 inet_gethost 4 rabbitmq 7769 7741 0 11:54 ? 00:00:00 /bin/sh /usr/lib/rabbitmq/bin/rabbitmq-server -noinput rabbitmq 7794 7769 0 11:54 ? 00:00:00 /bin/sh /usr/lib/rabbitmq/bin/rabbitmq-server -noinput rabbitmq 7795 7794 0 11:54 ? 00:00:00 ps -ef rabbitmq 7796 7794 0 11:54 ? 00:00:00 grep rabbit -bash-3.2# echo $$ 3834 

Read more

• From the shell (PID = 3834), run /etc/init.d/rabbitmq-server start;
• In the init.d script (PID = 7732) there is an upload (source) conf. file / etc / sysconfig / rabbitmq;
• From the init.d script, the shell script is run su rabbitmq -s / bin / sh -c / usr / lib / rabbitmq / bin / rabbitmq-multi "start_all" "1" (PID = 7740); (Source) / usr / lib / rabbitmq / bin / rabbitmq-evn is loaded from it, in which /etc/rabbimq/rabbitmq.conf is podguzhaetsya
• In / usr / lib / rabbitmq / bin / rabbitmq-multi, erlang (PID = 7741) is started with some parameter set. The most interesting is -s rabbit_multi. Specifying the function to be called instead of start (man erl);
• The shell script / bin / sh / usr / lib / rabbitmq / bin / rabbitmq-server -noinput (PID = 7769) is launched from the erlang application that is running erlang, in which the rabbitmq-server itself is launched.

Here is the code from rabbit_mulri.erl, in which the shell script is called:

 220 run_rabbitmq_server() -> 221 with_os([{unix, fun run_rabbitmq_server_unix/0}, 222 {win32, fun run_rabbitmq_server_win32/0}]). 223 224 run_rabbitmq_server_unix() -> 225 CmdLine = getenv("RABBITMQ_SCRIPT_HOME") ++ "/rabbitmq-server -noinput", 226 erlang:open_port({spawn, CmdLine}, [nouse_stdio]). 227 228 run_rabbitmq_server_win32() -> 229 Cmd = filename:nativename(os:find_executable("cmd")), 230 CmdLine = "\"" ++ getenv("RABBITMQ_SCRIPT_HOME") 231 ++ "\\rabbitmq-server.bat\" -noinput", 232 erlang:open_port({spawn_executable, Cmd}, 233 [{arg0, Cmd}, {args, ["/q", "/s", "/c", CmdLine]}, 234 nouse_stdio, hide]). 

Such a non-trivial launch process is clearly done because of the multiplatform nature of the application itself.

Environment variables

Now we have an understanding of how the rabbitmq-server launch process takes place, but there is no understanding how to influence it. Let's take a closer look at launching the erlang applications themselves

Launch launcher (rabbit_multi)

 exec erl \ -pa "${RABBITMQ_HOME}/ebin" \ -noinput \ -hidden \ ${RABBITMQ_MULTI_ERL_ARGS} \ -sname rabbitmq_multi$$ \ -s rabbit_multi \ ${RABBITMQ_MULTI_START_ARGS} \ -extra "$@" 

Server startup

 exec erl ${RABBITMQ_EBIN_PATH} ${RABBITMQ_START_RABBIT} -sname ${RABBITMQ_NODENAME} -boot ${RABBITMQ_BOOT_FILE} ${RABBITMQ_CONFIG_ARG} +W w \ ${RABBITMQ_SERVER_ERL_ARGS} \ ${RABBITMQ_LISTEN_ARG} \ -sasl errlog_type error \ -kernel error_logger '{file,"'${RABBITMQ_LOGS}'"}' \ -sasl sasl_error_logger '{file,"'${RABBITMQ_SASL_LOGS}'"}' \ -os_mon start_cpu_sup true \ -os_mon start_disksup false \ -os_mon start_memsup false \ -mnesia dir "\"${RABBITMQ_MNESIA_DIR}\"" \ ${RABBITMQ_SERVER_START_ARGS} \ "$@" 

I think you can see now which environment variables affect each of the erlang processes. It is recommended to change these parameters in /etc/rabbitmq/rabbitmq.conf, which, as we remember, is loaded when you start /usr/lib/rabbitmq/bin/rabbitmq-multi. This is how the started service with default parameters looks like:

 exec erl -noinput -sname rabbit@hostname -boot /var/lib/rabbitmq/mnesia/rabbit@hostname/plugins-scratch/rabbit \ -config /etc/rabbitmq/rabbitmq +W w +K true +A30 +P 1048576 -kernel inet_default_listen_options [{nodelay,true}] \ -kernel inet_default_connect_options [{nodelay,true}] \ -sasl errlog_type error -kernel error_logger '{file,'/var/log/rabbitmq/rabbit@hostname.log'}' \ -sasl sasl_error_logger '{file,'/var/log/rabbitmq/rabbit@hostname-sasl.log'}' \ -os_mon start_cpu_sup true -os_mon start_disksup false -os_mon start_memsup false -mnesia dir "/var/lib/rabbitmq/mnesia/rabbit@hostname" -noinput 

Total conf. files

As a result, 3 conf. Are responsible for rabbitmq-server operation. file:

• / etc / sysconfig / rabbitmq; - It is customary to indicate any low-level parameters (for example, ulimit)
• /etc/rabbitmq/rabbitmq.conf; - Configure environment variables to start erlang
• /etc/rabbitmq/rabbitmq.config. - It is configured itself rabbitmq-server

I hope someone this article will save some time. What values ​​to expose to variables affecting the operation of the server is a personal matter. Maybe I will write about this a little later.