docker-elk VS pfelk

Hello everyone! I am trying to get started with ELK and I am facing a very frustrating situation. I am trying to use the stack with Docker Compose. I have tried 2 versions: https://github.com/deviantony/docker-elk and also https://www.elastic.co/blog/getting-started-with-the-elastic-stack-and-docker-compose but they both have the same problem. I inspected the Docker container logs and I get some weird errors:

The waters are further muddied since I started out trying to spin up a docker instance https://github.com/deviantony/docker-elk but I found the config for docker is setup with a different layout, for example with logstash there is no conf.d directory, and pipelines are layed out differently, making it more challenging to use web examples. Overall I've tried many config examples and all have failed.

This is the github project conainer I'm trying to create a stack with: https://github.com/deviantony/docker-elk

Here's a quick way to get your hands into an elasticstack using docker-compose: https://github.com/deviantony/docker-elk

Essentially I've got 2 sets of standard JSON files that I'm trying to ingest into a dockerized ELK stack. The first set was downloaded cloudflare logs, standard line separated JSON data, tried to use Filebeat to ingest and it kept prefixing the JSON data with some dumb ECS event data, basically exactly what this post describes if its easier to see in pics. All of the cloudflare data was nested within the event.original field and would not get mapped. But once I tried to use just logstash directly, it was fine and mapped correctly and no more event data.

i use the elk into the docker,using the docker-elk compose , the logstash logs shows that it is reciving the logs:

It depends on what you want to get out of visualizing your logs. I use the combination of Elastic + Logstash + Kibana (ELK Stack) on docker to visualize things like

I did end up kinda Frankensteining this project and docker-elk. Basically took out the entire etc/pfelk directory from pfelk project and added the pipelines/dashboard/groks etc to docker-elk. This works really will for me since I have several other devices that aren’t OPNSense that I wanted ingested to ELK.

I understood that OPNsense runs fine with 8GB RAM and a relatively weak CPU, but then I saw this, which provides extended search and visualisation features to help you use the data created by OPNsense, and it recommends 32GB. pfelk/pfelk: pfSense/OPNsense + Elastic Stack (github.com)

I've carefully followed the manual Ubuntu setup of pfelk from https://github.com/pfelk/pfelk/tree/main, the instructions are pretty good. I did everything manually except for the dashboards, used the handy script. I've also configured syslog to remote log everything to it, plus unbound data.

I've been using pfElk. You could probably use some of the parsers from there to parse things yourself in Splunk.

On the left is a Kibana dashboard, showing information from the firewall (blocks/passes, connection type, etc). I use pfelk and customised the dashboard and the indexing a bit to suit my needs.

> So many chinese and russians IPs...

And Korean, and Dutch, I recall significant from Central America.

For anyone interested in what Geo's appear to be attacking you, and is a noob like me: https://github.com/pfelk/pfelk is really cool.