Skip to main content

This is for internal use by the PaaS team. Public-facing documentation is located at

Enhancing Kibana

We may find the need, to parse some more logs into Kibana for later use at convinient times.

How it works

The logstash config file is being generated:

  1. The logsearch-4cf packaging runs the build script which executes rake build.
  2. The logsearch-4cf rake build renders the main template file and creates logstash-filters-default.conf. It follows the parsing-rules order to acomplish it’s task.
  3. Logsearch functionality logstash_parser.filters, takes a list of files that must be present on disk. These contain logstash filters.
  4. Standard logsearch-4cf logstash-filters-default.conf is added to logstash_parser.filters.
  5. We add the path of our own custom-filters file to logstash_parser.filters.
  6. With help of our PR we can add content to the custom-filters file by filling in propertylogstash_parser.custom_filters
  7. The logsearch parser_ctl gathers all the files, including the ones from logstash_parser.filters to create /var/vcap/jobs/parser/config/logstash.conf which then is used by Logstash on each run.

Running Logstash

  1. Download logstash@5.0.0 - you will need Java 8.
  2. Install the translate plugin: logstash-plugin install logstash-filter-translate
  3. Copy /var/vcap/jobs/parser/config/logstash.conf locally or use the minimal config logsearch_logstash.conf
  4. Edit the config file and point to the log input file
  5. Run logstash: logstash -f ../../logstash/logsearch_logstash.conf
  6. You may find Grok debugger useful


Working with the above, may be a very unappealing process… If you combine it with you can save a lot of time

  1. To restart reading from the beginning of file: rm data/plugins/inputs/file/.sincedb_* and restart logstash
  2. To start reading the log file from beginning everytime you restart logstash:

    file {
        path => "/Users/colin/Documents/Boulot/gds/logstash/nginx_access.log"
        start_position => "beginning"
        sincedb_path => "/dev/null"
  3. If you combine it with you can save a lot of time

Working example

Now, let’s say we’ve got a log like:

May 11 13:24:09 57n27d4b-60p0-42ly-a00a-f913j7b8841f custom_nginx_access: - - [11/May/2017:13:24:09 +0000] "GET /healthz HTTP/1.1" 200 3 "-" "ELB-HealthChecker/1.0"

By adding few extra lines to our logstash configuration file, we can specify, what these actually are:

if [@source][component] == "custom_nginx_access" {
  grok {
    match => {
      "@message" =>
      '%{IPORHOST:[nginx][clientip]} - - \[%{HTTPDATE:[nginx][timestamp]}\] "%{WORD:[nginx][verb]} %{URIPATHPARAM:[nginx][request]} HTTP/%{NUMBER:[nginx][httpversion]}" %{NUMBER:[nginx][response]} (?:%{NUMBER:[nginx][bytes]}|-) (?:"(?:%{URI:[nginx][referrer]}|-)"|%{QS:[nginx][referrer]}) %{QS:[nginx][agent]}'

After a restart and successful entry in Kibana, it will look like that:

Key Value
nginx.agent “ELB-HealthChecker/1.0”
nginx.bytes 566
nginx.httpversion 1.1
nginx.request /info
nginx.response 200
nginx.response_time 0.029
nginx.timestamp 11/May/2017:14:59:04 +0000
nginx.vcap_request_id 9ed66476-764d-486e-b52c-05280929f726
nginx.verb GET