Running elasticsearch container on raspberry pi 3

Question

I am trying to run an containerized Elasticsearch of ELK stack on rpi3. There is no existing image on docker hub, so I am trying to build one from scratch. I used Dockerfile as base and change the base image to resin/rpi-raspbian. However, I encountered several problems:

1. For ES 5.4.0+, ES process failed to start up due to missing libjnidispatch.so in elastic-bundled jar file (Error: Native library (com/sun/jna/linux-arm/libjnidispatch.so) not found in resource path)
2. Therefore, I downgrade to se v5.3.2 instead. I got around jna problem and but ES still failed to bootstrap because due to runtime exception (seccomp unavailable: 'arm' architecture unsupported)

Elasticsearch log follows below. [2017-10-03T13:05:44,366][WARN ][o.e.b.JNANatives ] unable to install syscall filter: java.lang.UnsupportedOperationException: seccomp unavailable: 'arm' architecture unsupported

P.S. I already reduce heap size to 512m.

Is there anything I should do to make this happen? Thanks.

Answer 1

Answering my own question in case someone also stumble on the same problem. However, any other answers are welcome!

I came across with a git repo that builds a docker image running on rpi boards. It seems the author has modified the original elasticsearch source code which adds seccomp support on ARM. The docker repo for the modified distribution can be found here.

I have ElasticSearch 2.4 successfullly running on my raspberry pi 3 board. Here is my git repo for reference