Support #437
Updated by Daniel Curtis almost 10 years ago
I've decided to centralize all the logs generated by a client's production systems to a syslog server and after assessing a bunch of products, I am not working with Fluentd. The platform chosen to run Fluentd is FreeBSD inside a Jail (9.2-RELEASE at the time), a rock-solid and very well documented UNIX-like operating system. Besides, it also ships a production-ready ZFS implementation which always comes handy in the data center. FreeBSD-9.2 currently has the _sysutils/rubygem-fluentd_ is available in the ports tree,- tree, however the package has been marked as BROKEN-, BROKEN, and is not available using pkg. So I will document how to install Fluentd with ElasticSearch and Kibana. h2. Prerequisites * Install rubygem-fluentd <pre> pkg install rubygem-fluentd </pre> #* (*Deprecated*) In order to build Fluentd, there are are a few packages needed: <pre> pkg install git ruby ruby19-gems rubygem-bundler </pre> h3. (*Deprecated*) h2. Installation of Fluentd * Fetch the Fluentd source code: <pre> cd /usr/local git clone https://github.com/fluent/fluentd.git cd fluentd </pre> * Build and Install <pre> bundle install </pre> > Fetching gem metadata from https://rubygems.org/......... > ... > Your bundle is complete! > Use `bundle show [gemname]` to see where a bundled gem is installed. <pre> bundle exec rake build </pre> > fluentd xxx built to pkg/fluentd-xxx.gem. <pre> gem install pkg/fluentd-xxx.gem </pre> h3. Test Run * Run the following commands to to confirm that Fluentd was installed successfully: <pre> fluentd --setup ./fluent fluentd -c ./fluent/fluent.conf -vv & echo '{"json":"message"}' | fluent-cat debug.test </pre> #* The last command sends Fluentd a message ‘{“json”:”message”}’ with a “debug.test” tag. If the installation was successful, Fluentd will output the following message: <pre> > 2011-07-10 16:49:50 +0900 debug.test: {"json":"message"} </pre> The last command sends Fluentd a message ‘{“json”:”message”}’ with a “debug.test” tag. #* Kill the test run: <pre> fg 1 [ctrl]+[c] </pre> * Now enable fluentd to start at boot: <pre> echo 'fluentd_enable="YES"' >> /etc/rc.conf </pre> --- h2. Install ElasticSearch * Install ElasticSearch: <pre> pkg install elasticsearch </pre> #* Enable ElasticSearch at boot <pre> echo 'elasticsearch_enable="YES"' >> /etc/rc.conf </pre> #* Start ElasticSearch <pre> service elasticsearch start </pre> h3. Securing Elasticsearch * Up to version 1.2, Elasticsearch's dynamic scripting capability was enabled by default. Since this tutorial sets up the Kibana dashboard to be accessed from the public Internet, let's disable dynamic scripting by appending the following line at the end of the ElasticSearch configuration file: <pre> vi /usr/local/etc/elasticsearch/elasticsearch.yml </pre> > script.disable_dynamic: true * Restart Elasticsearch: <pre> service elasticsearch restart </pre> h2. Installing and Configuring Kibana h3. Getting Kibana * Change to the fluentd user, move to your home directory and download Kibana as follows: <pre> cd /usr/local wget http://download.elasticsearch.org/kibana/kibana/kibana-latest.zip unzip kibana-latest.zip mv kibana-latest kibana cd kibana </pre> h3. Configuring Kibana Since Kibana will use port 80 to talk to Elasticsearch as opposed to the default port 9200, Kibana's config.js must be updated. Open Kibana configuration file and look for the following line: <pre> vi config.js </pre> > elasticsearch: "http://"+window.location.hostname+":9200", and *replace it* with the following line: > elasticsearch: "http://"+window.location.hostname+":80", h2. Installing and Configuring Nginx (Proxy Server) We will use Nginx as a proxy server to allow access to the dashboard from the Public Internet (with basic authentication). * Install Nginx as follows: <pre> pkg install nginx py27-htpasswd </pre> #* Enable nginx to start at boot <pre> echo 'nginx_enable="YES"' >> /etc/rc.conf </pre> #* Start nginx <pre> service nginx start </pre> * Edit /usr/local/etc/nginx/nginx.conf and change the primary server block as follows: <pre> # # Nginx proxy for Elasticsearch + Kibana # # In this setup, we are password protecting the saving of dashboards. You may # wish to extend the password protection to all paths. # # Even though these paths are being called as the result of an ajax request, the # browser will prompt for a username/password on the first request # # If you use this, you'll want to point config.js at http://FQDN:80/ instead of # http://FQDN:9200 # server { listen *:80 ; server_name localhost; access_log /var/log/nginx-kibana.log; location / { root /usr/local/kibana; index index.html index.htm; } location ~ ^/_aliases$ { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; } location ~ ^/.*/_aliases$ { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; } location ~ ^/_nodes$ { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; } location ~ ^/.*/_search$ { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; } location ~ ^/.*/_mapping { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; } # Password protected end points location ~ ^/kibana-int/dashboard/.*$ { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; limit_except GET { proxy_pass http://127.0.0.1:9200; auth_basic "Restricted"; auth_basic_user_file /usr/local/etc/nginx/log.altservice.com.htpasswd; } } location ~ ^/kibana-int/temp.*$ { proxy_pass http://127.0.0.1:9200; proxy_read_timeout 90; limit_except GET { proxy_pass http://127.0.0.1:9200; auth_basic "Restricted"; auth_basic_user_file /usr/local/etc/nginx/log.altservice.com.htpasswd; } } </pre> * And generate a htpasswd file: python2.7 /usr/local/bin/htpasswd.py -c -b /usr/local/etc/nginx/log.altservice.com.htpasswd username SuperSecretPassword NOTE: Make sure to change the username and SuperSecretPassword to your needs * Finally, restart nginx as follows: <pre> service nginx restart </pre> Now, you should be able to see the generic Kibana dashboard at your server's IP address or domain, using your favorite browser. h2. Installing Fluentd Plugins We need a couple of plugins: # *out_elasticsearch*: this plugin lets Fluentd to stream data to Elasticsearch. # *outrecordreformer*: this plugin lets us process data into a more useful format. * The following commands install both plugins, it requires libcurl <pre> pkg install curl fpc-libcurl fluent-gem install fluent-plugin-elasticsearch fluent-gem install fluent-plugin-record-reformer </pre> h3. Add the Syslog configuration to Fluentd * Next, we configure Fluentd to listen to syslog messages and send them to Elasticsearch. Open /usr/local/fluentd/fluent.conf and add the following lines at the top of the file: <pre> ## Syslog input <source> type syslog port 5140 tag system </source> <match system.*.*> type record_reformer tag elasticsearch facility ${tag_parts[1]} severity ${tag_parts[2]} </match> <match elasticsearch> type copy <store> type stdout </store> <store> type elasticsearch logstash_format true flush_interval 5s #debug </store> </match> </pre> h2. Starting Fluentd * Start Fluentd with the following command: <pre> fluentd -c /usr/local/fluentd/fluent.conf -vv & </pre> h2. Forwarding Debian rsyslog Traffic to Fluentd I use Debian an many production systems, and one one the packages is rsyslogd. It needs to be reconfigured to forward syslog events to the port Fluentd listens to (port 5140 in this example). * Open the rsyslog configuration file and add the following line at the top <pre> sudo vi/etc/rsyslog.conf </pre> <pre> *.* @127.0.0.1:5140 </pre> * After saving and exiting the editor, restart rsyslogd as follows: <pre> sudo service rsyslog restart </pre> h2. Setting Up Kibana Dashboard Panels Kibana's default panels are very generic, so it's recommended to customize them. Here, we show two methods. h3. *Method 1*: Using a Template * The Fluentd team offers an alternative Kibana configuration that works with this setup better than the default one. To use this alternative configuration, run the following command: <pre> sudo cp default.json /usr/local/kibana/app/dashboards/default.json </pre> Note: The original configuration file is from the author's GitHub gist. If you refresh your Kibana dashboard home page at your server's URL, Kibana should now be configured to show histograms by syslog severity and facility, as well as recent log lines in a table. h3. *Method 2*: Manually Configuring Go to your server's IP address or domain to view the Kibana dashboard. There are a couple of starter templates, but let's choose the blank one called *Blank Dashboard*. Next, click on the *+ ADD A ROW* button on the right side of the dashboard. A configuration screen for a new row (a row consists of one or more panels) should show up. Enter a title, press the *Create Row* button, followed by *Save*. This creates a row. When an empty row is created, Kibana shows the prompt Add panel to empty row on the left. Click this button. It takes you to the configuration screen to add a new panel. Choose histogram from the dropdown menu. There are many parameters to configure for a new histogram, but you can just scroll down and press the Save button. This creates a new panel. h2. Resources http://docs.fluentd.org/articles/install-from-source https://www.digitalocean.com/community/tutorials/elasticsearch-fluentd-and-kibana-open-source-log-search-and-visualization