聚合存储在ElasticSearch中的Jenkins构建日志

Question

I'm storing my Jenkins build logs in ElasticSearch with the Jenkins Logstash plugin.

My configuration looks sort of like this:

That part works great, but I'd like to view the full log in Kibana.

The plugin incrementally sends the results to ES and breaks on each newline. That means a long log can look something like this in Kibana:

Where each line is a massive JSON output containing tons of fields I do not care about. I really only care about the message field.

I'm reading about aggregators right now that appear to be what I need, but my results are not coming out to what I'd like.

curl -X GET "localhost:9200/_search" -H 'Content-Type: application/json' -d'
{
    "aggs" : {
        "buildLog" : {
            "terms" : {
                "field" : "data.url"
            }
        }
    }
}'

Prints out a large glob of json that does not have what I need.

In a perfect world, I'd like to concatenate every message field from each data.url and fetch that.

In SQL, an individual query for this might look something like:

SELECT message FROM jenkins-logstash WHERE data.url='job/playground/36' ORDER BY ASC

Where 'job/playground/36' is one example of every data.url.

How can I go about doing this?

Answer 1

Update: Better answer than before.

I still ended up using FileBeat, but with ELK v6.5.+ Kibana has a logs UI! https://www.elastic.co/guide/en/kibana/current/logs-ui.html

The default config from FileBeat works fine with it.

__

Old answer:

I ended up solving this by using FileBeat to harvest all logs, and then using the Kibana Log Viewer to watch each one. I filtered based on source and then used the path where the log was going to be.