# Rabbit MQ setup and MaxScale Integration ## Introduction A step by step guide helps installing a RabbitMQ server and testing it before MaxScale integration. New plugin filter and a message consumer application need to be compiled and linked with an external C library, RabbitMQ-c, that provides AMQP protocol integration. Custom configuration, with TCP/IP and Queue parameters, is also detailed here. The software install setup provides RPM and DEB packaging and traditional compilation steps. ## Step 1 - Get the RabbitMQ binaries On Centos 6.5 using fedora / RHEL rpm get the rpm from [http://www.rabbitmq.com/](http://www.rabbitmq.com/ "RabbitMQ") rabbitmq-server-3.3.4-1.noarch.rpm Please note, before installing RabbitMQ, you must install Erlang. Example: yum install erlang Package erlang-R14B-04.3.el6.x86_64 already installed and latest version ## Step 2 - Install and Start the Server Install the packages using your distribution's package manager and start the server: yum install rabbitmq-server-3.3.4-1.noarch.rpm systemctl start rabbitmq-server.service To configure your RabbitMQ server, please refer to the RabbitMQ website: [http://www.rabbitmq.com/](http://www.rabbitmq.com/ RabbitMQ website). rabbitmqctl is a command line tool for managing a RabbitMQ broker. It performs all actions by connecting to one of the broker's nodes. rabbitmqctl list_queues rabbitmqctl list_queues | list_exchanges| cluster_status | list_bindings | list_connections | list_consumers | status Example output: [root@maxscale-02 MaxScale]# rabbitmqctl status Status of node 'rabbit@maxscale-02' ... [{pid,12251}, {running_applications,[{rabbit,"RabbitMQ","3.3.4"}, {os_mon,"CPO CXC 138 46","2.2.7"}, {xmerl,"XML parser","1.2.10"}, {mnesia,"MNESIA CXC 138 12","4.5"}, {sasl,"SASL CXC 138 11","2.1.10"}, {stdlib,"ERTS CXC 138 10","1.17.5"}, {kernel,"ERTS CXC 138 10","2.14.5"}]}, {os,{unix,linux}}, {erlang_version,"Erlang R14B04 (erts-5.8.5) [source] [64-bit] [smp:2:2] [rq:2] [async-threads:30] [kernel-poll:true]\n"}, ... {listeners,[{clustering,25672,"::"},{amqp,5672,"::"}]}, ... ...done. [root@maxscale-02 MaxScale]# rabbitmqctl list_bindings Listing bindings ... x1 exchange q1 queue k1 [] ...done. Interaction with the server may require stop & reset at some point: rabbitmqctl stop_app rabbitmqctl reset rabbitmqctl start_app ## Step 3 - Install and test the client libraries The selected library for MaxScale integration of RabbitMQ is: [https://github.com/alanxz/rabbitmq-c](https://github.com/alanxz/rabbitmq-c RabbitMQ-C) ### Manual software compilation To compile the RabbitMQ-C libraries manually: git clone https://github.com/alanxz/rabbitmq-c.git cd rabbitmq-c cmake -DCMAKE_INSTALL_PREFIX=/usr . make make install Please note, this will install the packages to /usr. If you do not wish to install them to this location, provide a different value for the CMAKE_INSTALL_PREFIX variable. ### Setup using the EPEL repository Check how to configure your distribution for the EPEL repository: [https://fedoraproject.org/wiki/EPEL](https://fedoraproject.org/wiki/EPEL EPEL) Configure your repositories and install the software: yum install librabbitmq.x86_64 you might also like to install: librabbitmq-tools.x86_64, librabbitmq-devel.x86_64 Please note you may also install the rabbitmq server from the EPEL repository: yum install rabbitmq-server ### Basic tests with library The required library librabbitmq-c is now installed and we continue with basic operations with amqp_* tools, located in the examples/ folder of the build directory, testing client server interaction. Please note, those example applications may not be included in the RPM library packages. #### Test 1 - create the exchange [root@maxscale-02 examples]# ./amqp_exchange_declare Usage: amqp_exchange_declare host port exchange exchangetype Declare the exchange: [root@maxscale-02 examples]# ./amqp_exchange_declare 127.0.0.1 5672 foo direct #### Test 2 - Listen to exchange with selected binding key [root@maxscale-02 examples]# ./amqp_listen Usage: amqp_listen host port exchange bindingkey Start the listener: [root@maxscale-02 examples]# ./amqp_listen 127.0.0.1 5672 foo k1 & #### Test 3 - Send a message … [root@maxscale-02 examples]# ./amqp_sendstring Usage: amqp_sendstring host port exchange routingkey messagebody [root@maxscale-02 examples]# ./amqp_sendstring 127.0.0.1 5672 foo k1 “This is a new message” ... and watch the listener output Delivery 1, exchange foo routingkey k1 Content-type: text/plain ## Step 4 - MaxScale integration with librabbitmq-c A new filter (mqfilter.c) is implemented in order to send messages to the rabbitmq server and a message consumer (rabbitmq_consumer/consumer.c) program will get messages and store them into a MySQL/MariaDB database. A quick way to install MaxScale with the RabbitMQ filter is to go to the MaxScale source directory and run the following commands: mkdir build cd build cmake .. -DBUILD_RABBITMQ=Y make make install To build the RabbitMQ filter CMake needs an additional parameter: -DBUILD_RABBITMQ=Y If the librabbitmq-c library is manually compiled it may be necessary to manually pass the location of the libraries and header files to CMake. Libraries: -DRABBITMQ_LIBRARIES= Headers: -DRABBITMQ_HEADERS= Please note, Message Queue Consumer (consumer.c) also needs to be compiled with MySQL/MariaDB client libraries in addition to the RabbitMQ-c libraries. If you have your MySQL/MariaDB client libraries and headers in non-standard locations, you can pass them manually to CMake: Libraries: -DMYSQLCLIENT_LIBRARIES= Headers: -DMYSQLCLIENT_HEADERS= The message queue consumer must be also built as a separate task, it’s not built as part of MaxScale build system. To build it, run the following commands in the rabbitmq_consumer directory in the MaxScale source folder: mkdir build cd build cmake .. make To install it: make install To build packages: make package This generates RPM or DEB packages based on your system. These packages can then be installed on remote systems for easy access to the data generated by the consumer client. ## Step 5 - Configure new applications The new filter needs to be configured in MaxScale.cnf. [Test Service] type=service router=readconnroute router_options=slave servers=server1,server2,server3,server5,server4 user=massi passwd=massi filters=MQ [MQ] type=filter module=mqfilter exchange=x1 key=k1 queue=q1 hostname=127.0.0.1 port=5672 logging_trigger=all Logging triggers define whether to log all or a subset of the incoming queries using these options: # log only some elements or all logging_trigger=[all,source,schema,object] # Whether to log only SELECT, UPDATE, INSERT and DELETE queries or all possible queries logging_log_all=true|false # Log only when any of the trigger parameters match or only if all parameters match logging_strict=true|false # specify objects logging_object=mytable,another_table # specify logged users logging_source_user=testuser,testuser # specify source addresses logging_source_host=127.0.0.1,192.168.10.14 # specify schemas logging_schema=employees,orders,catalog Example: logging_trigger=object,schema,source logging_strict=false logging_log_all=false logging_object=my1 logging_schema=test logging_source_user=maxtest The logging result of the example is: if user maxtest does something, it's logged and all queries in test schema are logged anything targeting my1 table is logged SELECT NOW(), SELECT MD5(“xyz)” are not logged Please note that if we want to log only the user ‘maxtest’ accessing the schema ‘test’ with target ‘my1’ the option logging_strict must be set to TRUE and if we want to include those selects without schema name the option logging_log_all must be set to TRUE. The mqfilter logs into the MaxScale TRACE log information about the matched logging triggers and the message delivering: 2014 09/03 06:22:04 Trigger is TRG_SOURCE: user: testuser = testuser 2014 09/03 06:22:04 Trigger is TRG_SCHEMA: test = test 2014 09/03 06:22:04 Trigger is TRG_OBJECT: test.t1 = t1 2014 09/03 06:22:04 Routing message to: 127.0.0.1:5672 / as guest/guest, exchange: x1 key:k1 queue:q1 The consumer application needs to be configured as well: #The options for the consumer are: #hostname RabbitMQ hostname #port RabbitMQ port #vhost RabbitMQ virtual host #user RabbitMQ username #passwd RabbitMQ password #queue Name of the queue to use #dbserver SQL server name #dbport SQL server port #dbname Name of the database to use #dbuser SQL server username #dbpasswd SQL server password #logfile Message log filename [consumer] hostname=127.0.0.1 port=5672 vhost=/ user=guest passwd=guest queue=q1 dbserver=127.0.0.1 dbport=3308 dbname=mqpairs dbuser=xxx dbpasswd=yyy We may probably need to modify LD_LIBRARY_PATH before launching ‘consumer’: # export LD_LIBRARY_PATH=/packages/rabbitmq-c/rabbitmq-c/librabbitmq:/packages/mariadb_client-2.0.0-Linux/lib/mariadb:/usr/lib64 and finally we can launch it: # ./consumer If the consumer.cnf file is not in the same directory as the binary file is, you can provide the location of the folder that it is in by passing it the -c flag followed by the path: # ./consumer -c path/to/file and start maxScale as well ## Step 6 - Test the filter and check collected data Assuming that MaxScale and the message consumer are successfully running let’s connect to the service with an active mqfilter: [root@maxscale-02 MaxScale]# mysql -h 127.0.0.1 -P 4506 -uxxx -pyyy ... MariaDB [(none)]> select RAND(3), RAND(5); +--------------------+---------------------+ | RAND(3) | RAND(5) | +--------------------+---------------------+ | 0.9057697559760601 | 0.40613597483014313 | +--------------------+---------------------+ 1 row in set (0.01 sec) … MariaDB [(none)]> select RAND(3544), RAND(11); we can check the consumer output in the terminal where it was started: -------------------------------------------------------------- Received: 1409671452|select @@version_comment limit ? Received: 1409671452|Columns: 1 ... Received: 1409671477|select RAND(?), RAND(?) Received: 1409671477|Columns: 2 We query now the database for the content collected so far: MariaDB [(none)]> use mqpairs; Database changed MariaDB [mqpairs]> select * from pairs; +-------------------------------------+----------------------------------+------------+---------------------+---------------------+---------+ | tag | query | reply | date_in | date_out | counter | +-------------------------------------+----------------------------------+------------+---------------------+---------------------+---------+ | 006c006d006e006f007000710072007374 | select @@version_comment limit ? | Columns: 1 | 2014-09-02 11:14:51 | 2014-09-02 11:26:38 | 3 | | 00750076007700780079007a007b007c7d | SELECT DATABASE() | Columns: 1 | 2014-09-02 11:14:56 | 2014-09-02 11:27:06 | 3 | | 007e007f00800081008200830084008586 | show databases | Columns: 1 | 2014-09-02 11:14:56 | 2014-09-02 11:27:06 | 3 | | 008700880089008a008b008c008d008e8f | show tables | Columns: 1 | 2014-09-02 11:14:56 | 2014-09-02 11:27:06 | 3 | | 0090009100920093009400950096009798 | select * from mqpairs.pairs | Columns: 6 | 2014-09-02 11:15:00 | 2014-09-02 11:27:00 | 12 | | 00fc00fd00fe00ff0100010101020103104 | select NOW() | Columns: 1 | 2014-09-02 11:24:23 | 2014-09-02 11:24:23 | 1 | | 01050106010701080109010a010b010c10d | select RAND(?), RAND(?) | Columns: 2 | 2014-09-02 11:24:37 | 2014-09-02 11:24:37 | 1 | +-------------------------------------+----------------------------------+------------+---------------------+---------------------+---------+ 7 rows in set (0.01 sec) The filter send queries to the RabbitMQ server in the canonical format, i.e select RAND(?), RAND(?). The queries Message Queue Consumer application gets from the server are stored with a counter that quickly shows how many times that normalized query was received: | 01050106010701080109010a010b010c10d | select RAND(?), RAND(?) | Columns: 2 | 2014-09-02 11:24:37 | 2014-09-02 11:29:15 | 3 |