Hello!
I'm running nginx in a docker container on kubernetes, and my log lines look like this:
{"log":"10.72.0.1 - - [05/Nov/2016:02:05:38 +0000] \"GET / HTTP/1.1\" 304 0 \"-\" \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2906.0 Safari/537.36\" \"-\"\n","stream":"stdout","tag":"kubernetes.var.log.containers.my-nginx-379829228-ouolf_default_my-nginx-409eb594dc73fe7a8fd95e102d83fe7ccb5e42c6c7c99f7000516f773939ec3a.log","@timestamp":"2016-11-05T02:05:38+00:00"}
I currently have fluentd running in the default setup and it is doing a good job of pulling out the different keys in this json, but the actual nginx log all comes out as a single string. I would like to parse the "log" value, which is from nginx, so that elasticsearch can e.g. bucket on response status or URI.
Is this recommendable? Do I need a mixture of plugins for something like this? I am completely new to fluentd and I'm not sure where to look for advice on this. Thanks for any help!
Riley